abs
stringlengths 44
64
| Download PDF
stringlengths 75
115
| OpenReview
stringlengths 42
42
| title
stringlengths 15
148
| url
stringlengths 44
64
| authors
stringlengths 6
903
| detail_url
stringlengths 44
64
| tags
stringclasses 1
value | abstract
stringlengths 422
5.84k
|
|---|---|---|---|---|---|---|---|---|
https://proceedings.mlr.press/v235/quintas-martinez24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/quintas-martinez24a/quintas-martinez24a.pdf
|
https://openreview.net/forum?id=n2eppIzHlL
|
Multiply-Robust Causal Change Attribution
|
https://proceedings.mlr.press/v235/quintas-martinez24a.html
|
Victor Quintas-Martinez, Mohammad Taha Bahadori, Eduardo Santiago, Jeff Mu, David Heckerman
|
https://proceedings.mlr.press/v235/quintas-martinez24a.html
|
ICML 2024
|
Comparing two samples of data, we observe a change in the distribution of an outcome variable. In the presence of multiple explanatory variables, how much of the change can be explained by each possible cause? We develop a new estimation strategy that, given a causal model, combines regression and re-weighting methods to quantify the contribution of each causal mechanism. Our proposed methodology is multiply robust, meaning that it still recovers the target parameter under partial misspecification. We prove that our estimator is consistent and asymptotically normal. Moreover, it can be incorporated into existing frameworks for causal attribution, such as Shapley values, which will inherit the consistency and large-sample distribution properties. Our method demonstrates excellent performance in Monte Carlo simulations, and we show its usefulness in an empirical application. Our method is implemented as part of the Python library “DoWhy“ (Sharma & Kiciman, 2020; Blöbaum et al., 2022).
|
https://proceedings.mlr.press/v235/rafiey24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rafiey24a/rafiey24a.pdf
|
https://openreview.net/forum?id=SAbZExIIgG
|
Decomposable Submodular Maximization in Federated Setting
|
https://proceedings.mlr.press/v235/rafiey24a.html
|
Akbar Rafiey
|
https://proceedings.mlr.press/v235/rafiey24a.html
|
ICML 2024
|
Submodular functions, as well as the sub-class of decomposable submodular functions, and their optimization appear in a wide range of applications in machine learning, recommendation systems, and welfare maximization. However, optimization of decomposable submodular functions with millions of component functions is computationally prohibitive. Furthermore, the component functions may be private (they might represent user preference function, for example) and cannot be widely shared. To address these issues, we propose a federated optimization setting for decomposable submodular optimization. In this setting, clients have their own preference functions, and a weighted sum of these preferences needs to be maximized. We implement the popular continuous greedy algorithm in this setting where clients take parallel small local steps towards the local solution and then the local changes are aggregated at a central server. To address the large number of clients, the aggregation is performed only on a subsampled set. Further, the aggregation is performed only intermittently between stretches of parallel local steps, which reduces communication cost significantly. We show that our federated algorithm is guaranteed to provide a good approximate solution, even in the presence of above cost-cutting measures. Finally, we show how the federated setting can be incorporated in solving fundamental discrete submodular optimization problems such as Maximum Coverage and Facility Location.
|
https://proceedings.mlr.press/v235/raghvendra24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/raghvendra24a/raghvendra24a.pdf
|
https://openreview.net/forum?id=opieUcKjPa
|
A New Robust Partial p-Wasserstein-Based Metric for Comparing Distributions
|
https://proceedings.mlr.press/v235/raghvendra24a.html
|
Sharath Raghvendra, Pouyan Shirzadian, Kaiyi Zhang
|
https://proceedings.mlr.press/v235/raghvendra24a.html
|
ICML 2024
|
The $2$-Wasserstein distance is sensitive to minor geometric differences between distributions, making it a very powerful dissimilarity metric. However, due to this sensitivity, a small outlier mass can also cause a significant increase in the $2$-Wasserstein distance between two similar distributions. Similarly, sampling discrepancy can cause the empirical $2$-Wasserstein distance on $n$ samples in $\mathbb{R}^2$ to converge to the true distance at a rate of $n^{-1/4}$, which is significantly slower than the rate of $n^{-1/2}$ for $1$-Wasserstein distance. We introduce a new family of distances parameterized by $k \ge 0$, called $k$-RPW that is based on computing the partial $2$-Wasserstein distance. We show that (1) $k$-RPW satisfies the metric properties, (2) $k$-RPW is robust to small outlier mass while retaining the sensitivity of $2$-Wasserstein distance to minor geometric differences, and (3) when $k$ is a constant, $k$-RPW distance between empirical distributions on $n$ samples in $\mathbb{R}^2$ converges to the true distance at a rate of $n^{-1/3}$, which is faster than the convergence rate of $n^{-1/4}$ for the $2$-Wasserstein distance. Using the partial $p$-Wasserstein distance, we extend our distance to any $p \in [1,\infty]$. By setting parameters $k$ or $p$ appropriately, we can reduce our distance to the total variation, $p$-Wasserstein, and the Lévy-Prokhorov distances. Experiments show that our distance function achieves higher accuracy in comparison to the $1$-Wasserstein, $2$-Wasserstein, and TV distances for image retrieval tasks on noisy real-world data sets.
|
https://proceedings.mlr.press/v235/rahman24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rahman24a/rahman24a.pdf
|
https://openreview.net/forum?id=bOhzU7NpTB
|
Modular Learning of Deep Causal Generative Models for High-dimensional Causal Inference
|
https://proceedings.mlr.press/v235/rahman24a.html
|
Md Musfiqur Rahman, Murat Kocaoglu
|
https://proceedings.mlr.press/v235/rahman24a.html
|
ICML 2024
|
Sound and complete algorithms have been proposed to compute identifiable causal queries using the causal structure and data. However, most of these algorithms assume accurate estimation of the data distribution, which is impractical for high-dimensional variables such as images. On the other hand, modern deep generative architectures can be trained to sample from high-dimensional distributions. However, training these networks are typically very costly. Thus, it is desirable to leverage pre-trained models to answer causal queries using such high-dimensional data. To address this, we propose modular training of deep causal generative models that not only makes learning more efficient, but also allows us to utilize large, pre-trained conditional generative models. To the best of our knowledge, our algorithm, Modular-DCM is the first algorithm that, given the causal structure, uses adversarial training to learn the network weights, and can make use of pre-trained models to provably sample from any identifiable causal query in the presence of latent confounders. With extensive experiments on the Colored-MNIST dataset, we demonstrate that our algorithm outperforms the baselines. We also show our algorithm’s convergence on the COVIDx dataset and its utility with a causal invariant prediction problem on CelebA-HQ.
|
https://proceedings.mlr.press/v235/rahmani24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rahmani24a/rahmani24a.pdf
|
https://openreview.net/forum?id=biE1uHyG0l
|
Fundamental Limits of Distributed Covariance Matrix Estimation Under Communication Constraints
|
https://proceedings.mlr.press/v235/rahmani24a.html
|
Mohammad Reza Rahmani, Mohammad Hossein Yassaee, Mohammad Ali Maddah-Ali, Mohammad Reza Aref
|
https://proceedings.mlr.press/v235/rahmani24a.html
|
ICML 2024
|
Estimating high-dimensional covariance matrices is crucial in various domains. This work considers a scenario where two collaborating agents access disjoint dimensions of $m$ samples from a high–dimensional random vector, and they can only communicate a limited number of bits to a central server, which wants to accurately approximate the covariance matrix. We analyze the fundamental trade–off between communication cost, number of samples, and estimation accuracy. We prove a lower bound on the error achievable by any estimator, highlighting the impact of dimensions, number of samples, and communication budget. Furthermore, we present an algorithm that achieves this lower bound up to a logarithmic factor, demonstrating its near-optimality in practical settings.
|
https://proceedings.mlr.press/v235/raisa24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/raisa24a/raisa24a.pdf
|
https://openreview.net/forum?id=gTBjkJvadC
|
Subsampling is not Magic: Why Large Batch Sizes Work for Differentially Private Stochastic Optimisation
|
https://proceedings.mlr.press/v235/raisa24a.html
|
Ossi Räisä, Joonas Jälkö, Antti Honkela
|
https://proceedings.mlr.press/v235/raisa24a.html
|
ICML 2024
|
We study how the batch size affects the total gradient variance in differentially private stochastic gradient descent (DP-SGD), seeking a theoretical explanation for the usefulness of large batch sizes. As DP-SGD is the basis of modern DP deep learning, its properties have been widely studied, and recent works have empirically found large batch sizes to be beneficial. However, theoretical explanations of this benefit are currently heuristic at best. We first observe that the total gradient variance in DP-SGD can be decomposed into subsampling-induced and noise-induced variances. We then prove that in the limit of an infinite number of iterations, the effective noise-induced variance is invariant to the batch size. The remaining subsampling-induced variance decreases with larger batch sizes, so large batches reduce the effective total gradient variance. We confirm numerically that the asymptotic regime is relevant in practical settings when the batch size is not small, and find that outside the asymptotic regime, the total gradient variance decreases even more with large batch sizes. We also find a sufficient condition that implies that large batch sizes similarly reduce effective DP noise variance for one iteration of DP-SGD.
|
https://proceedings.mlr.press/v235/rakotoarison24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rakotoarison24a/rakotoarison24a.pdf
|
https://openreview.net/forum?id=VyoY3Wh9Wd
|
In-Context Freeze-Thaw Bayesian Optimization for Hyperparameter Optimization
|
https://proceedings.mlr.press/v235/rakotoarison24a.html
|
Herilalaina Rakotoarison, Steven Adriaensen, Neeratyoy Mallik, Samir Garibov, Eddie Bergman, Frank Hutter
|
https://proceedings.mlr.press/v235/rakotoarison24a.html
|
ICML 2024
|
With the increasing computational costs associated with deep learning, automated hyperparameter optimization methods, strongly relying on black-box Bayesian optimization (BO), face limitations. Freeze-thaw BO offers a promising grey-box alternative, strategically allocating scarce resources incrementally to different configurations. However, the frequent surrogate model updates inherent to this approach pose challenges for existing methods, requiring retraining or fine-tuning their neural network surrogates online, introducing overhead, instability, and hyper-hyperparameters. In this work, we propose FT-PFN, a novel surrogate for Freeze-thaw style BO. FT-PFN is a prior-data fitted network (PFN) that leverages the transformers’ in-context learning ability to efficiently and reliably do Bayesian learning curve extrapolation in a single forward pass. Our empirical analysis across three benchmark suites shows that the predictions made by FT-PFN are more accurate and 10-100 times faster than those of the deep Gaussian process and deep ensemble surrogates used in previous work. Furthermore, we show that, when combined with our novel acquisition mechanism (MFPI-random), the resulting in-context freeze-thaw BO method (ifBO), yields new state-of-the-art performance in the same three families of deep learning HPO benchmarks considered in prior work.
|
https://proceedings.mlr.press/v235/raman24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/raman24a/raman24a.pdf
|
https://openreview.net/forum?id=JA6ThxAmth
|
Understanding Inter-Concept Relationships in Concept-Based Models
|
https://proceedings.mlr.press/v235/raman24a.html
|
Naveen Janaki Raman, Mateo Espinosa Zarlenga, Mateja Jamnik
|
https://proceedings.mlr.press/v235/raman24a.html
|
ICML 2024
|
Concept-based explainability methods provide insight into deep learning systems by constructing explanations using human-understandable concepts. While the literature on human reasoning demonstrates that we exploit relationships between concepts when solving tasks, it is unclear whether concept-based methods incorporate the rich structure of inter-concept relationships. We analyse the concept representations learnt by concept-based models to understand whether these models correctly capture inter-concept relationships. First, we empirically demonstrate that state-of-the-art concept-based models produce representations that lack stability and robustness, and such methods fail to capture inter-concept relationships. Then, we develop a novel algorithm which leverages inter-concept relationships to improve concept intervention accuracy, demonstrating how correctly capturing inter-concept relationships can improve downstream tasks.
|
https://proceedings.mlr.press/v235/raman24b.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/raman24b/raman24b.pdf
|
https://openreview.net/forum?id=nU1mtFDtMX
|
STEER: Assessing the Economic Rationality of Large Language Models
|
https://proceedings.mlr.press/v235/raman24b.html
|
Narun Krishnamurthi Raman, Taylor Lundy, Samuel Joseph Amouyal, Yoav Levine, Kevin Leyton-Brown, Moshe Tennenholtz
|
https://proceedings.mlr.press/v235/raman24b.html
|
ICML 2024
|
There is increasing interest in using LLMs as decision-making "agents". Doing so includes many degrees of freedom: which model should be used; how should it be prompted; should it be asked to introspect, conduct chain-of-thought reasoning, etc? Settling these questions—and more broadly, determining whether an LLM agent is reliable enough to be trusted—requires a methodology for assessing such an agent’s economic rationality. In this paper, we provide one. We begin by surveying the economic literature on rational decision making, taxonomizing a large set of fine-grained "elements" that an agent should exhibit, along with dependencies between them. We then propose a benchmark distribution that quantitatively scores an LLMs performance on these elements and, combined with a user-provided rubric, produces a "rationality report card". Finally, we describe the results of a large-scale empirical experiment with 14 different LLMs, characterizing the both current state of the art and the impact of different model sizes on models’ ability to exhibit rational behavior.
|
https://proceedings.mlr.press/v235/rame24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rame24a/rame24a.pdf
|
https://openreview.net/forum?id=s7RDnNUJy6
|
WARM: On the Benefits of Weight Averaged Reward Models
|
https://proceedings.mlr.press/v235/rame24a.html
|
Alexandre Rame, Nino Vieillard, Leonard Hussenot, Robert Dadashi-Tazehozi, Geoffrey Cideron, Olivier Bachem, Johan Ferret
|
https://proceedings.mlr.press/v235/rame24a.html
|
ICML 2024
|
Aligning large language models (LLMs) with human preferences through reinforcement learning (RLHF) can lead to reward hacking, where LLMs exploit failures in the reward model (RM) to achieve seemingly high rewards without meeting the underlying objectives. We identify two primary challenges when designing RMs to mitigate reward hacking: distribution shifts during the RL process and inconsistencies in human preferences. As a solution, we propose Weight Averaged Reward Models (WARM), first fine-tuning multiple RMs, then averaging them in the weight space. This strategy follows the observation that fine-tuned weights remain linearly mode connected when sharing the same pre-training. By averaging weights, WARM improves efficiency compared to the traditional ensembling of predictions, while improving reliability under distribution shifts and robustness to preference inconsistencies. Our experiments on summarization tasks, using best-of-N and RL methods, shows that WARM improves the overall quality and alignment of LLM predictions; for example, a policy RL fine-tuned with WARM has a 79.4% win rate against a policy RL fine-tuned with a single RM.
|
https://proceedings.mlr.press/v235/ramesh24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/ramesh24a/ramesh24a.pdf
|
https://openreview.net/forum?id=L1eJ3NKPCd
|
Compositional Capabilities of Autoregressive Transformers: A Study on Synthetic, Interpretable Tasks
|
https://proceedings.mlr.press/v235/ramesh24a.html
|
Rahul Ramesh, Ekdeep Singh Lubana, Mikail Khona, Robert P. Dick, Hidenori Tanaka
|
https://proceedings.mlr.press/v235/ramesh24a.html
|
ICML 2024
|
Transformers trained on huge text corpora exhibit a remarkable set of capabilities, e.g., performing simple logical operations. Given the inherent compositional nature of language, one can expect the model to learn to compose these capabilities, potentially yielding a combinatorial explosion of what operations it can perform on an input. Motivated by the above, we aim to assess in this paper “how capable can a transformer become?”. Specifically, we train autoregressive Transformer models on a data-generating process that involves compositions of a set of well-defined monolithic capabilities. Through a series of extensive and systematic experiments on this data-generating process, we show that: (1) autoregressive Transformers can learn compositional structures from small amounts of training data and generalize to exponentially or even combinatorially many functions; (2) composing functions by generating intermediate outputs is more effective at generalizing to unseen compositions, compared to generating no intermediate outputs; (3) biases in the order of the compositions in the training data, results in Transformers that fail to compose some combinations of functions; and (4) the attention layers seem to select the capability to apply while the feed-forward layers execute the capability.
|
https://proceedings.mlr.press/v235/ramesh24b.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/ramesh24b/ramesh24b.pdf
|
https://openreview.net/forum?id=NlM4gp8hyO
|
Sequence Compression Speeds Up Credit Assignment in Reinforcement Learning
|
https://proceedings.mlr.press/v235/ramesh24b.html
|
Aditya Ramesh, Kenny John Young, Louis Kirsch, Jürgen Schmidhuber
|
https://proceedings.mlr.press/v235/ramesh24b.html
|
ICML 2024
|
Temporal credit assignment in reinforcement learning is challenging due to delayed and stochastic outcomes. Monte Carlo targets can bridge long delays between action and consequence but lead to high-variance targets due to stochasticity. Temporal difference (TD) learning uses bootstrapping to overcome variance but introduces a bias that can only be corrected through many iterations. TD($\lambda$) provides a mechanism to navigate this bias-variance tradeoff smoothly. Appropriately selecting $\lambda$ can significantly improve performance. Here, we propose Chunked-TD, which uses predicted probabilities of transitions from a model for computing $\lambda$-return targets. Unlike other model-based solutions to credit assignment, Chunked-TD is less vulnerable to model inaccuracies. Our approach is motivated by the principle of history compression and ‘chunks’ trajectories for conventional TD learning. Chunking with learned world models compresses near-deterministic regions of the environment-policy interaction to speed up credit assignment while still bootstrapping when necessary. We propose algorithms that can be implemented online and show that they solve some problems much faster than conventional TD($\lambda$).
|
https://proceedings.mlr.press/v235/rane24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rane24a/rane24a.pdf
|
https://openreview.net/forum?id=PQWVUbqQtQ
|
Position: The Reasonable Person Standard for AI
|
https://proceedings.mlr.press/v235/rane24a.html
|
Sunayana Rane
|
https://proceedings.mlr.press/v235/rane24a.html
|
ICML 2024
|
As AI systems are increasingly incorporated into domains where human behavior has set the norm, a challenge for AI governance and AI alignment research is to regulate their behavior in a way that is useful and constructive for society. One way to answer this question is to ask: how do we govern the human behavior that the models are emulating? To evaluate human behavior, the American legal system often uses the "Reasonable Person Standard." The idea of "reasonable" behavior comes up in nearly every area of law. The legal system often judges the actions of parties with respect to what a reasonable person would have done under similar circumstances. This paper argues that the reasonable person standard provides useful guidelines for the type of behavior we should develop, probe, and stress-test in models. It explains how reasonableness is defined and used in key areas of the law using illustrative cases, how the reasonable person standard could apply to AI behavior in each of these areas and contexts, and how our societal understanding of "reasonable" behavior provides useful technical goals for AI researchers.
|
https://proceedings.mlr.press/v235/raparthy24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/raparthy24a/raparthy24a.pdf
|
https://openreview.net/forum?id=lVQ4FUZ6dp
|
Generalization to New Sequential Decision Making Tasks with In-Context Learning
|
https://proceedings.mlr.press/v235/raparthy24a.html
|
Sharath Chandra Raparthy, Eric Hambro, Robert Kirk, Mikael Henaff, Roberta Raileanu
|
https://proceedings.mlr.press/v235/raparthy24a.html
|
ICML 2024
|
Training autonomous agents that can learn new tasks from only a handful of demonstrations is a long-standing problem in machine learning. Recently, transformers have been shown to learn new language or vision tasks without any weight updates from only a few examples, also referred to as in-context learning. However, the sequential decision making setting poses additional challenges having a lower tolerance for errors since the environment’s stochasticity or the agent’s actions can lead to unseen, and sometimes unrecoverable, states. In this paper, we use an illustrative example to show that naively applying transformers to sequential decision making problems does not enable in-context learning of new tasks. We then demonstrate how training on sequences of trajectories with certain distributional properties leads to in-context learning of new sequential decision making tasks. We investigate different design choices and find that larger model and dataset sizes, as well as more task diversity, environment stochasticity, and trajectory burstiness, all result in better in-context learning of new out-of-distribution tasks. By training on large diverse offline datasets, our model is able to learn new MiniHack and Procgen tasks without any weight updates from just a handful of demonstrations.
|
https://proceedings.mlr.press/v235/rathore24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rathore24a/rathore24a.pdf
|
https://openreview.net/forum?id=mJGiFr8jLa
|
Challenges in Training PINNs: A Loss Landscape Perspective
|
https://proceedings.mlr.press/v235/rathore24a.html
|
Pratik Rathore, Weimu Lei, Zachary Frangella, Lu Lu, Madeleine Udell
|
https://proceedings.mlr.press/v235/rathore24a.html
|
ICML 2024
|
This paper explores challenges in training Physics-Informed Neural Networks (PINNs), emphasizing the role of the loss landscape in the training process. We examine difficulties in minimizing the PINN loss function, particularly due to ill-conditioning caused by differential operators in the residual term. We compare gradient-based optimizers Adam, L-BFGS, and their combination Adam+L-BFGS, showing the superiority of Adam+L-BFGS, and introduce a novel second-order optimizer, NysNewton-CG (NNCG), which significantly improves PINN performance. Theoretically, our work elucidates the connection between ill-conditioned differential operators and ill-conditioning in the PINN loss and shows the benefits of combining first- and second-order optimization methods. Our work presents valuable insights and more powerful optimization strategies for training PINNs, which could improve the utility of PINNs for solving difficult partial differential equations.
|
https://proceedings.mlr.press/v235/ravikumar24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/ravikumar24a/ravikumar24a.pdf
|
https://openreview.net/forum?id=4dxR7awO5n
|
Unveiling Privacy, Memorization, and Input Curvature Links
|
https://proceedings.mlr.press/v235/ravikumar24a.html
|
Deepak Ravikumar, Efstathia Soufleri, Abolfazl Hashemi, Kaushik Roy
|
https://proceedings.mlr.press/v235/ravikumar24a.html
|
ICML 2024
|
Deep Neural Nets (DNNs) have become a pervasive tool for solving many emerging problems. However, they tend to overfit to and memorize the training set. Memorization is of keen interest since it is closely related to several concepts such as generalization, noisy learning, and privacy. To study memorization, Feldman (2019) proposed a formal score, however its computational requirements limit its practical use. Recent research has shown empirical evidence linking input loss curvature (measured by the trace of the loss Hessian w.r.t inputs) and memorization. It was shown to be $\sim3$ orders of magnitude more efficient than calculating the memorization score. However, there is a lack of theoretical understanding linking memorization with input loss curvature. In this paper, we not only investigate this connection but also extend our analysis to establish theoretical links between differential privacy, memorization, and input loss curvature. First, we derive an upper bound on memorization characterized by both differential privacy and input loss curvature. Secondly, we present a novel insight showing that input loss curvature is upper-bounded by the differential privacy parameter. Our theoretical findings are further validated using deep models on CIFAR and ImageNet datasets, showing a strong correlation between our theoretical predictions and results observed in practice.
|
https://proceedings.mlr.press/v235/rawal24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rawal24a/rawal24a.pdf
|
https://openreview.net/forum?id=Wj5wm3Os5v
|
Dissecting Multimodality in VideoQA Transformer Models by Impairing Modality Fusion
|
https://proceedings.mlr.press/v235/rawal24a.html
|
Ishaan Singh Rawal, Alexander Matyasko, Shantanu Jaiswal, Basura Fernando, Cheston Tan
|
https://proceedings.mlr.press/v235/rawal24a.html
|
ICML 2024
|
While VideoQA Transformer models demonstrate competitive performance on standard benchmarks, the reasons behind their success are not fully understood. Do these models capture the rich multimodal structures and dynamics from video and text jointly? Or are they achieving high scores by exploiting biases and spurious features? Hence, to provide insights, we design QUAG (QUadrant AveraGe), a lightweight and non-parametric probe, to conduct dataset-model combined representation analysis by impairing modality fusion. We find that the models achieve high performance on many datasets without leveraging multimodal representations. To validate QUAG further, we design QUAG-attention, a less-expressive replacement of self-attention with restricted token interactions. Models with QUAG-attention achieve similar performance with significantly fewer multiplication operations without any finetuning. Our findings raise doubts about the current models’ abilities to learn highly-coupled multimodal representations. Hence, we design the CLAVI (Complements in LAnguage and VIdeo) dataset, a stress-test dataset curated by augmenting real-world videos to have high modality coupling. Consistent with the findings of QUAG, we find that most of the models achieve near-trivial performance on CLAVI. This reasserts the limitations of current models for learning highly-coupled multimodal representations, that is not evaluated by the current datasets.
|
https://proceedings.mlr.press/v235/ray-chaudhury24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/ray-chaudhury24a/ray-chaudhury24a.pdf
|
https://openreview.net/forum?id=6Zgjrowepn
|
Fair Federated Learning via the Proportional Veto Core
|
https://proceedings.mlr.press/v235/ray-chaudhury24a.html
|
Bhaskar Ray Chaudhury, Aniket Murhekar, Zhuowen Yuan, Bo Li, Ruta Mehta, Ariel D. Procaccia
|
https://proceedings.mlr.press/v235/ray-chaudhury24a.html
|
ICML 2024
|
Previous work on fairness in federated learning introduced the notion of core stability, which provides utility-based fairness guarantees to any subset of participating agents. However, these guarantees require strong assumptions on agent utilities that render them impractical. To address this shortcoming, we measure the quality of output models in terms of their ordinal rank instead of their cardinal utility, and use this insight to adapt the classical notion of proportional veto core (PVC) from social choice theory to the federated learning setting. We prove that models that are PVC-stable exist in very general learning paradigms, even allowing non-convex model sets, as well as non-convex and non-concave loss functions. We also design Rank-Core-Fed, a distributed federated learning algorithm, to train a PVC-stable model. Finally, we demonstrate that Rank-Core-Fed outperforms baselines in terms of fairness on different datasets.
|
https://proceedings.mlr.press/v235/ray-chowdhury24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/ray-chowdhury24a/ray-chowdhury24a.pdf
|
https://openreview.net/forum?id=yhpDKSw7yA
|
Provably Robust DPO: Aligning Language Models with Noisy Feedback
|
https://proceedings.mlr.press/v235/ray-chowdhury24a.html
|
Sayak Ray Chowdhury, Anush Kini, Nagarajan Natarajan
|
https://proceedings.mlr.press/v235/ray-chowdhury24a.html
|
ICML 2024
|
Learning from preference-based feedback has recently gained traction as a promising approach to align language models with human interests. While these aligned generative models have demonstrated impressive capabilities across various tasks, their dependence on high-quality human preference data poses a bottleneck in practical applications. Specifically, noisy (incorrect and ambiguous) preference pairs in the dataset might restrict the language models from capturing human intent accurately. While practitioners have recently proposed heuristics to mitigate the effect of noisy preferences, a complete theoretical understanding of their workings remain elusive. In this work, we aim to bridge this gap by introducing a general framework for policy optimization in the presence of random preference flips. We focus on the direct preference optimization (DPO) algorithm in particular since it assumes that preferences adhere to the Bradley-Terry-Luce (BTL) model, raising concerns about the impact of noisy data on the learned policy. We design a novel loss function, which de-bias the effect of noise on average, making a policy trained by minimizing that loss robust to the noise. Under log-linear parameterization of the policy class and assuming good feature coverage of the SFT policy, we prove that the sub-optimality gap of the proposed robust DPO (rDPO) policy compared to the optimal policy is of the order $O(\frac{1}{1-2\epsilon}\sqrt{\frac{d}{n}})$, where $\epsilon < 1/2$ is flip rate of labels, $d$ is policy parameter dimension and $n$ is size of dataset. Our experiments on IMDb sentiment generation and Anthropic’s helpful-harmless dataset shows that rDPO is robust to noise in preference labels compared to vanilla DPO and other heuristics proposed by practitioners.
|
https://proceedings.mlr.press/v235/razin24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/razin24a/razin24a.pdf
|
https://openreview.net/forum?id=XT6iF8FDZx
|
Implicit Bias of Policy Gradient in Linear Quadratic Control: Extrapolation to Unseen Initial States
|
https://proceedings.mlr.press/v235/razin24a.html
|
Noam Razin, Yotam Alexander, Edo Cohen-Karlik, Raja Giryes, Amir Globerson, Nadav Cohen
|
https://proceedings.mlr.press/v235/razin24a.html
|
ICML 2024
|
In modern machine learning, models can often fit training data in numerous ways, some of which perform well on unseen (test) data, while others do not. Remarkably, in such cases gradient descent frequently exhibits an implicit bias that leads to excellent performance on unseen data. This implicit bias was extensively studied in supervised learning, but is far less understood in optimal control (reinforcement learning). There, learning a controller applied to a system via gradient descent is known as policy gradient, and a question of prime importance is the extent to which a learned controller extrapolates to unseen initial states. This paper theoretically studies the implicit bias of policy gradient in terms of extrapolation to unseen initial states. Focusing on the fundamental Linear Quadratic Regulator (LQR) problem, we establish that the extent of extrapolation depends on the degree of exploration induced by the system when commencing from initial states included in training. Experiments corroborate our theory, and demonstrate its conclusions on problems beyond LQR, where systems are non-linear and controllers are neural networks. We hypothesize that real-world optimal control may be greatly improved by developing methods for informed selection of initial states to train on.
|
https://proceedings.mlr.press/v235/reifenstein24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/reifenstein24a/reifenstein24a.pdf
|
https://openreview.net/forum?id=BrCrnaCYDc
|
Dynamic Anisotropic Smoothing for Noisy Derivative-Free Optimization
|
https://proceedings.mlr.press/v235/reifenstein24a.html
|
Sam Reifenstein, Timothee Leleu, Yoshihisa Yamamoto
|
https://proceedings.mlr.press/v235/reifenstein24a.html
|
ICML 2024
|
We propose a novel algorithm that extends the methods of ball smoothing and Gaussian smoothing for noisy derivative-free optimization by accounting for the heterogeneous curvature of the objective function. The algorithm dynamically adapts the shape of the smoothing kernel to approximate the Hessian of the objective function around a local optimum. This approach significantly reduces the error in estimating the gradient from noisy evaluations through sampling. We demonstrate the efficacy of our method through numerical experiments on artificial problems. Additionally, we show improved performance when tuning NP-hard combinatorial optimization solvers compared to existing state-ofthe-art heuristic derivative-free and Bayesian optimization methods.
|
https://proceedings.mlr.press/v235/reizinger24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/reizinger24a/reizinger24a.pdf
|
https://openreview.net/forum?id=pVyOchWUBa
|
Position: Understanding LLMs Requires More Than Statistical Generalization
|
https://proceedings.mlr.press/v235/reizinger24a.html
|
Patrik Reizinger, Szilvia Ujváry, Anna Mészáros, Anna Kerekes, Wieland Brendel, Ferenc Huszár
|
https://proceedings.mlr.press/v235/reizinger24a.html
|
ICML 2024
|
The last decade has seen blossoming research in deep learning theory attempting to answer, “Why does deep learning generalize?" A powerful shift in perspective precipitated this progress: the study of overparametrized models in the interpolation regime. In this paper, we argue that another perspective shift is due, since some of the desirable qualities of LLMs are not a consequence of good statistical generalization and require a separate theoretical explanation. Our core argument relies on the observation that AR probabilistic models are inherently non-identifiable: models zero or near-zero KL divergence apart—thus, equivalent test loss—can exhibit markedly different behaviors. We support our position with mathematical examples and empirical observations, illustrating why non-identifiability has practical relevance through three case studies: (1) the non-identifiability of zero-shot rule extrapolation; (2) the approximate non-identifiability of in-context learning; and (3) the non-identifiability of fine-tunability. We review promising research directions focusing on LLM-relevant generalization measures, transferability, and inductive biases.
|
https://proceedings.mlr.press/v235/ren24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/ren24a/ren24a.pdf
|
https://openreview.net/forum?id=DM0r4qatjT
|
Optimal Batched Linear Bandits
|
https://proceedings.mlr.press/v235/ren24a.html
|
Xuanfei Ren, Tianyuan Jin, Pan Xu
|
https://proceedings.mlr.press/v235/ren24a.html
|
ICML 2024
|
We introduce the E$^4$ algorithm for the batched linear bandit problem, incorporating an Explore-Estimate-Eliminate-Exploit framework. With a proper choice of exploration rate, we prove E$^4$ achieves the finite-time minimax optimal regret with only $O(\log\log T)$ batches, and the asymptotically optimal regret with only $3$ batches as $T\rightarrow\infty$, where $T$ is the time horizon. We further prove a lower bound on the batch complexity of liner contextual bandits showing that any asymptotically optimal algorithm must require at least $3$ batches in expectation as $T\rightarrow \infty$, which indicates E$^4$ achieves the asymptotic optimality in regret and batch complexity simultaneously. To the best of our knowledge, E$^4$ is the first algorithm for linear bandits that simultaneously achieves the minimax and asymptotic optimality in regret with the corresponding optimal batch complexities. In addition, we show that with another choice of exploration rate E$^4$ achieves an instance-dependent regret bound requiring at most $O(\log T)$ batches, and maintains the minimax optimality and asymptotic optimality. We conduct thorough experiments to evaluate our algorithm on randomly generated instances and the challenging End of Optimism instances (Lattimore & Szepesvari, 2017) which were shown to be hard to learn for optimism based algorithms. Empirical results show that E$^4$ consistently outperforms baseline algorithms with respect to regret minimization, batch complexity, and computational efficiency.
|
https://proceedings.mlr.press/v235/ren24b.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/ren24b/ren24b.pdf
|
https://openreview.net/forum?id=LZeixIvQcB
|
TabLog: Test-Time Adaptation for Tabular Data Using Logic Rules
|
https://proceedings.mlr.press/v235/ren24b.html
|
Weijieying Ren, Xiaoting Li, Huiyuan Chen, Vineeth Rakesh, Zhuoyi Wang, Mahashweta Das, Vasant G Honavar
|
https://proceedings.mlr.press/v235/ren24b.html
|
ICML 2024
|
We consider the problem of test-time adaptation of predictive models trained on tabular data. Effective solution of this problem requires adaptation of predictive models trained on the source domain to a target domain, using only unlabeled target domain data, without access to source domain data. Existing test-time adaptation methods for tabular data have difficulty coping with the heterogeneous features and their complex dependencies inherent in tabular data. To overcome these limitations, we consider test-time adaptation in the setting wherein the logical structure of the rules is assumed to remain invariant despite distribution shift between source and target domains whereas the numerical parameters associated with the rules and the weights assigned to them can vary to accommodate distribution shift. TabLog discretizes numerical features, models dependencies between heterogeneous features, introduces a novel contrastive loss for coping with distribution shift, and presents an end-to-end framework for efficient training and test-time adaptation by taking advantage of a logical neural network representation of a rule ensemble. We present results of experiments using several benchmark data sets that demonstrate TabLog is competitive with or improves upon the state-of-the-art methods for test-time adaptation of predictive models trained on tabular data. Our code is available at https://github.com/WeijieyingRen/TabLog.
|
https://proceedings.mlr.press/v235/ren24c.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/ren24c/ren24c.pdf
|
https://openreview.net/forum?id=2zI2scD2Iz
|
Hybrid Inverse Reinforcement Learning
|
https://proceedings.mlr.press/v235/ren24c.html
|
Juntao Ren, Gokul Swamy, Steven Wu, Drew Bagnell, Sanjiban Choudhury
|
https://proceedings.mlr.press/v235/ren24c.html
|
ICML 2024
|
The inverse reinforcement learning approach to imitation learning is a double-edged sword. On the one hand, it can enable learning from a smaller number of expert demonstrations with more robustness to error compounding than behavioral cloning approaches. On the other hand, it requires that the learner repeatedly solve a computationally expensive reinforcement learning (RL) problem. Often, much of this computation is wasted searching over policies very dissimilar to the expert’s. In this work, we propose using hybrid RL – training on a mixture of online and expert data – to curtail unnecessary exploration. Intuitively, the expert data focuses the learner on good states during training, which reduces the amount of exploration required to compute a strong policy. Notably, such an approach doesn’t need the ability to reset the learner to arbitrary states in the environment, a requirement of prior work in efficient inverse RL. More formally, we derive a reduction from inverse RL to expert-competitive RL (rather than globally optimal RL) that allows us to dramatically reduce interaction during the inner policy search loop while maintaining the benefits of the IRL approach. This allows us to derive both model-free and model-based hybrid inverse RL algorithms with strong policy performance guarantees. Empirically, we find that our approaches are significantly more sample efficient than standard inverse RL and several other baselines on a suite of continuous control tasks.
|
https://proceedings.mlr.press/v235/ren24d.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/ren24d/ren24d.pdf
|
https://openreview.net/forum?id=mzGtunvpJH
|
Rejuvenating image-GPT as Strong Visual Representation Learners
|
https://proceedings.mlr.press/v235/ren24d.html
|
Sucheng Ren, Zeyu Wang, Hongru Zhu, Junfei Xiao, Alan Yuille, Cihang Xie
|
https://proceedings.mlr.press/v235/ren24d.html
|
ICML 2024
|
This paper enhances image-GPT (iGPT), one of the pioneering works that introduce autoregressive pretraining to predict the next pixels for visual representation learning. Two simple yet essential changes are made. First, we shift the prediction target from raw pixels to semantic tokens, enabling a higher-level understanding of visual content. Second, we supplement the autoregressive modeling by instructing the model to predict not only the next tokens but also the visible tokens. This pipeline is particularly effective when semantic tokens are encoded by discriminatively trained models, such as CLIP. We introduce this novel approach as D-iGPT. Extensive experiments showcase that D-iGPT excels as a strong learner of visual representations: A notable achievement is its compelling performance on the ImageNet-1K dataset — by training on publicly available datasets, D-iGPT unprecedentedly achieves 90.0% top-1 accuracy with a vanilla ViT-H. Additionally, D-iGPT shows strong generalization on the downstream task. Code is available at https://github.com/OliverRensu/D-iGPT.
|
https://proceedings.mlr.press/v235/ren24e.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/ren24e/ren24e.pdf
|
https://openreview.net/forum?id=FSxTEvuFa7
|
CarbonNovo: Joint Design of Protein Structure and Sequence Using a Unified Energy-based Model
|
https://proceedings.mlr.press/v235/ren24e.html
|
Milong Ren, Tian Zhu, Haicang Zhang
|
https://proceedings.mlr.press/v235/ren24e.html
|
ICML 2024
|
De novo protein design aims to create novel protein structures and sequences unseen in nature. Recent structure-oriented design methods typically employ a two-stage strategy, where structure design and sequence design modules are trained separately, and the backbone structures and sequences are generated sequentially in inference. While diffusion-based generative models like RFdiffusion show great promise in structure design, they face inherent limitations within the two-stage framework. First, the sequence design module risks overfitting, as the accuracy of the generated structures may not align with that of the crystal structures used for training. Second, the sequence design module lacks interaction with the structure design module to further optimize the generated structures. To address these challenges, we propose CarbonNovo, a unified energy-based model for jointly generating protein structure and sequence. Specifically, we leverage a score-based generative model and Markov Random Fields for describing the energy landscape of protein structure and sequence. In CarbonNovo, the structure and sequence design module communicates at each diffusion step, encouraging the generation of more coherent structure-sequence pairs. Moreover, the unified framework allows for incorporating the protein language models as evolutionary constraints for generated proteins. The rigorous evaluation demonstrates that CarbonNovo outperforms two-stage methods across various metrics, including designability, novelty, sequence plausibility, and Rosetta Energy.
|
https://proceedings.mlr.press/v235/renaud24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/renaud24a/renaud24a.pdf
|
https://openreview.net/forum?id=byAXJTk0LH
|
Plug-and-Play image restoration with Stochastic deNOising REgularization
|
https://proceedings.mlr.press/v235/renaud24a.html
|
Marien Renaud, Jean Prost, Arthur Leclaire, Nicolas Papadakis
|
https://proceedings.mlr.press/v235/renaud24a.html
|
ICML 2024
|
Plug-and-Play (PnP) algorithms are a class of iterative algorithms that address image inverse problems by combining a physical model and a deep neural network for regularization. Even if they produce impressive image restoration results, these algorithms rely on a non-standard use of a denoiser on images that are less and less noisy along the iterations, which contrasts with recent algorithms based on Diffusion Models (DM), where the denoiser is applied only on re-noised images. We propose a new PnP framework, called Stochastic deNOising REgularization (SNORE), which applies the denoiser only on images with noise of the adequate level. It is based on an explicit stochastic regularization, which leads to a stochastic gradient descent algorithm to solve ill-posed inverse problems. A convergence analysis of this algorithm and its annealing extension is provided. Experimentally, we prove that SNORE is competitive with respect to state-of-the-art methods on deblurring and inpainting tasks, both quantitatively and qualitatively.
|
https://proceedings.mlr.press/v235/reshef24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/reshef24a/reshef24a.pdf
|
https://openreview.net/forum?id=sTVSyqD6XX
|
Private and Federated Stochastic Convex Optimization: Efficient Strategies for Centralized Systems
|
https://proceedings.mlr.press/v235/reshef24a.html
|
Roie Reshef, Kfir Yehuda Levy
|
https://proceedings.mlr.press/v235/reshef24a.html
|
ICML 2024
|
This paper addresses the challenge of preserving privacy in Federated Learning (FL) within centralized systems, focusing on both trusted and untrusted server scenarios. We analyze this setting within the Stochastic Convex Optimization (SCO) framework, and devise methods that ensure Differential Privacy (DP) while maintaining optimal convergence rates for homogeneous and heterogeneous data distributions. Our approach, based on a recent stochastic optimization technique, offers linear computational complexity, comparable to non-private FL methods, and reduced gradient obfuscation. This work enhances the practicality of DP in FL, balancing privacy, efficiency, and robustness in a variety of server trust environments.
|
https://proceedings.mlr.press/v235/reuel24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/reuel24a/reuel24a.pdf
|
https://openreview.net/forum?id=Be2B6f0ps1
|
Position: Technical Research and Talent is Needed for Effective AI Governance
|
https://proceedings.mlr.press/v235/reuel24a.html
|
Anka Reuel, Lisa Soder, Benjamin Bucknall, Trond Arne Undheim
|
https://proceedings.mlr.press/v235/reuel24a.html
|
ICML 2024
|
In light of recent advancements in AI capabilities and the increasingly widespread integration of AI systems into society, governments worldwide are actively seeking to mitigate the potential harms and risks associated with these technologies through regulation and other governance tools. However, there exist significant gaps between governance aspirations and the current state of the technical tooling necessary for their realisation. In this position paper, we survey policy documents published by public-sector institutions in the EU, US, and China to highlight specific areas of disconnect between the technical requirements necessary for enacting proposed policy actions, and the current technical state of the art. Our analysis motivates a call for tighter integration of the AI/ML research community within AI governance in order to i) catalyse technical research aimed at bridging the gap between current and supposed technical underpinnings of regulatory action, as well as ii) increase the level of technical expertise within governing institutions so as to inform and guide effective governance of AI.
|
https://proceedings.mlr.press/v235/ribar24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/ribar24a/ribar24a.pdf
|
https://openreview.net/forum?id=OS5dqxmmtl
|
SparQ Attention: Bandwidth-Efficient LLM Inference
|
https://proceedings.mlr.press/v235/ribar24a.html
|
Luka Ribar, Ivan Chelombiev, Luke Hudlass-Galley, Charlie Blake, Carlo Luschi, Douglas Orr
|
https://proceedings.mlr.press/v235/ribar24a.html
|
ICML 2024
|
The computational difficulties of large language model (LLM) inference remain a significant obstacle to their widespread deployment. The need for many applications to support long input sequences and process them in large batches typically causes token-generation to be bottlenecked by data transfer. For this reason, we introduce SparQ Attention, a technique for increasing the inference throughput of LLMs by utilising memory bandwidth more efficiently within the attention layers, through selective fetching of the cached history. Our proposed technique can be applied directly to off-the-shelf LLMs during inference, without requiring any modification to the pre-training setup or additional fine-tuning. We show that SparQ Attention brings up to 8x savings in attention data transfers without substantial drops in accuracy, by evaluating Llama 2 and 3, Mistral, Gemma and Pythia models on a wide range of downstream tasks.
|
https://proceedings.mlr.press/v235/ringel24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/ringel24a/ringel24a.pdf
|
https://openreview.net/forum?id=0b7txvPYlr
|
Early Time Classification with Accumulated Accuracy Gap Control
|
https://proceedings.mlr.press/v235/ringel24a.html
|
Liran Ringel, Regev Cohen, Daniel Freedman, Michael Elad, Yaniv Romano
|
https://proceedings.mlr.press/v235/ringel24a.html
|
ICML 2024
|
Early time classification algorithms aim to label a stream of features without processing the full input stream, while maintaining accuracy comparable to that achieved by applying the classifier to the entire input. In this paper, we introduce a statistical framework that can be applied to any sequential classifier, formulating a calibrated stopping rule. This data-driven rule attains finite-sample, distribution-free control of the accuracy gap between full and early-time classification. We start by presenting a novel method that builds on the Learn-then-Test calibration framework to control this gap marginally, on average over i.i.d. instances. As this algorithm tends to yield an excessively high accuracy gap for early halt times, our main contribution is the proposal of a framework that controls a stronger notion of error, where the accuracy gap is controlled conditionally on the accumulated halt times. Numerical experiments demonstrate the effectiveness, applicability, and usefulness of our method. We show that our proposed early stopping mechanism reduces up to 94% of timesteps used for classification while achieving rigorous accuracy gap control.
|
https://proceedings.mlr.press/v235/robertson24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/robertson24a/robertson24a.pdf
|
https://openreview.net/forum?id=qklMNNub0H
|
Implicit Regularization in Feedback Alignment Learning Mechanisms for Neural Networks
|
https://proceedings.mlr.press/v235/robertson24a.html
|
Zachary Robertson, Sanmi Koyejo
|
https://proceedings.mlr.press/v235/robertson24a.html
|
ICML 2024
|
Feedback Alignment (FA) methods are biologically inspired local learning rules for training neural networks with reduced communication between layers. While FA has potential applications in distributed and privacy-aware ML, limitations in multi-class classification and lack of theoretical understanding of the alignment mechanism have constrained its impact. This study introduces a unified framework elucidating the operational principles behind alignment in FA. Our key contributions include: (1) a novel conservation law linking changes in synaptic weights to implicit regularization that maintains alignment with the gradient, with support from experiments, (2) sufficient conditions for convergence based on the concept of alignment dominance, and (3) empirical analysis showing better alignment can enhance FA performance on complex multi-class tasks. Overall, these theoretical and practical advancements improve interpretability of bio-plausible learning rules and provide groundwork for developing enhanced FA algorithms.
|
https://proceedings.mlr.press/v235/rodomanov24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rodomanov24a/rodomanov24a.pdf
|
https://openreview.net/forum?id=Wnhp34K5jR
|
Universal Gradient Methods for Stochastic Convex Optimization
|
https://proceedings.mlr.press/v235/rodomanov24a.html
|
Anton Rodomanov, Ali Kavis, Yongtao Wu, Kimon Antonakopoulos, Volkan Cevher
|
https://proceedings.mlr.press/v235/rodomanov24a.html
|
ICML 2024
|
We develop universal gradient methods for Stochastic Convex Optimization (SCO). Our algorithms automatically adapt not only to the oracle’s noise but also to the Hölder smoothness of the objective function without a priori knowledge of the particular setting. The key ingredient is a novel strategy for adjusting step-size coefficients in the Stochastic Gradient Method (SGD). Unlike AdaGrad, which accumulates gradient norms, our Universal Gradient Method accumulates appropriate combinations of gradientand iterate differences. The resulting algorithm has state-of-the-art worst-case convergence rate guarantees for the entire Hölder class including, in particular, both nonsmooth functions and those with Lipschitz continuous gradient. We also present the Universal Fast Gradient Method for SCO enjoying optimal efficiency estimates.
|
https://proceedings.mlr.press/v235/rogers24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rogers24a/rogers24a.pdf
|
https://openreview.net/forum?id=M2cwkGleRL
|
Position: Key Claims in LLM Research Have a Long Tail of Footnotes
|
https://proceedings.mlr.press/v235/rogers24a.html
|
Anna Rogers, Sasha Luccioni
|
https://proceedings.mlr.press/v235/rogers24a.html
|
ICML 2024
|
Much of the recent discourse within the ML community has been centered around Large Language Models (LLMs), their functionality and potential – yet not only do we not have a working definition of LLMs, but much of this discourse relies on claims and assumptions that are worth re-examining. We contribute a definition of LLMs, critically examine five common claims regarding their properties (including ’emergent properties’), and conclude with suggestions for future research directions and their framing.
|
https://proceedings.mlr.press/v235/roh24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/roh24a/roh24a.pdf
|
https://openreview.net/forum?id=4ZrppmS42b
|
LEVI: Generalizable Fine-tuning via Layer-wise Ensemble of Different Views
|
https://proceedings.mlr.press/v235/roh24a.html
|
Yuji Roh, Qingyun Liu, Huan Gui, Zhe Yuan, Yujin Tang, Steven Euijong Whang, Liang Liu, Shuchao Bi, Lichan Hong, Ed H. Chi, Zhe Zhao
|
https://proceedings.mlr.press/v235/roh24a.html
|
ICML 2024
|
Fine-tuning is becoming widely used for leveraging the power of pre-trained foundation models in new downstream tasks. While there are many successes of fine-tuning on various tasks, recent studies have observed challenges in the generalization of fine-tuned models to unseen distributions (i.e., out-of-distribution; OOD). To improve OOD generalization, some previous studies identify the limitations of fine-tuning data and regulate fine-tuning to preserve the general representation learned from pre-training data. However, potential limitations in the pre-training data and models are often ignored. In this paper, we contend that overly relying on the pre-trained representation may hinder fine-tuning from learning essential representations for downstream tasks and thus hurt its OOD generalization. It can be especially catastrophic when new tasks are from different (sub)domains compared to pre-training data. To address the issues in both pre-training and fine-tuning data, we propose a novel generalizable fine-tuning method LEVI (Layer-wise Ensemble of different VIews), where the pre-trained model is adaptively ensembled layer-wise with a small task-specific model, while preserving its efficiencies. By combining two complementing models, LEVI effectively suppresses problematic features in both the fine-tuning data and pre-trained model and preserves useful features for new tasks. Broad experiments with large language and vision models show that LEVI greatly improves fine-tuning generalization via emphasizing different views from fine-tuning data and pre-trained features.
|
https://proceedings.mlr.press/v235/rolf24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rolf24a/rolf24a.pdf
|
https://openreview.net/forum?id=PQ0ERKKYJu
|
Position: Mission Critical – Satellite Data is a Distinct Modality in Machine Learning
|
https://proceedings.mlr.press/v235/rolf24a.html
|
Esther Rolf, Konstantin Klemmer, Caleb Robinson, Hannah Kerner
|
https://proceedings.mlr.press/v235/rolf24a.html
|
ICML 2024
|
Satellite data has the potential to inspire a seismic shift for machine learning—one in which we rethink existing practices designed for traditional data modalities. As machine learning for satellite data (SatML) gains traction for its real-world impact, our field is at a crossroads. We can either continue applying ill-suited approaches, or we can initiate a new research agenda that centers around the unique characteristics and challenges of satellite data. This position paper argues that satellite data constitutes a distinct modality for machine learning research and that we must recognize it as such to advance the quality and impact of SatML research across theory, methods, and deployment. We outline research directions, critical discussion questions and actionable suggestions to transform SatML from merely an intriguing application area to a dedicated research discipline that helps move the needle on big challenges for machine learning and society.
|
https://proceedings.mlr.press/v235/rolnick24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rolnick24a/rolnick24a.pdf
|
https://openreview.net/forum?id=xEB2oF3vvb
|
Position: Application-Driven Innovation in Machine Learning
|
https://proceedings.mlr.press/v235/rolnick24a.html
|
David Rolnick, Alan Aspuru-Guzik, Sara Beery, Bistra Dilkina, Priya L. Donti, Marzyeh Ghassemi, Hannah Kerner, Claire Monteleoni, Esther Rolf, Milind Tambe, Adam White
|
https://proceedings.mlr.press/v235/rolnick24a.html
|
ICML 2024
|
In this position paper, we argue that application-driven research has been systemically under-valued in the machine learning community. As applications of machine learning proliferate, innovative algorithms inspired by specific real-world challenges have become increasingly important. Such work offers the potential for significant impact not merely in domains of application but also in machine learning itself. In this paper, we describe the paradigm of application-driven research in machine learning, contrasting it with the more standard paradigm of methods-driven research. We illustrate the benefits of application-driven machine learning and how this approach can productively synergize with methods-driven work. Despite these benefits, we find that reviewing, hiring, and teaching practices in machine learning often hold back application-driven innovation. We outline how these processes may be improved.
|
https://proceedings.mlr.press/v235/rosenfeld24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rosenfeld24a/rosenfeld24a.pdf
|
https://openreview.net/forum?id=OURP5Z58jt
|
One-Shot Strategic Classification Under Unknown Costs
|
https://proceedings.mlr.press/v235/rosenfeld24a.html
|
Elan Rosenfeld, Nir Rosenfeld
|
https://proceedings.mlr.press/v235/rosenfeld24a.html
|
ICML 2024
|
The goal of strategic classification is to learn decision rules which are robust to strategic input manipulation. Earlier works assume that these responses are known; while some recent works handle unknown responses, they exclusively study online settings with repeated model deployments. But there are many domains – particularly in public policy, a common motivating use case – where multiple deployments are infeasible, or where even one bad round is unacceptable. To address this gap, we initiate the formal study of one-shot strategic classification under unknown responses, which requires committing to a single classifier once. Focusing on uncertainty in the users’ cost function, we begin by proving that for a broad class of costs, even a small mis-estimation of the true cost can entail trivial accuracy in the worst case. In light of this, we frame the task as a minimax problem, aiming to minimize worst-case risk over an uncertainty set of costs. We design efficient algorithms for both the full-batch and stochastic settings, which we prove converge (offline) to the minimax solution at the rate of $\tilde{\mathcal{O}}(T^{-\frac{1}{2}})$. Our analysis reveals important structure stemming from strategic responses, particularly the value of dual norm regularization with respect to the cost function.
|
https://proceedings.mlr.press/v235/ruaud24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/ruaud24a/ruaud24a.pdf
|
https://openreview.net/forum?id=vBJZ93tvoE
|
Modelling Microbial Communities with Graph Neural Networks
|
https://proceedings.mlr.press/v235/ruaud24a.html
|
Albane Ruaud, Cansu Sancaktar, Marco Bagatella, Christoph Ratzke, Georg Martius
|
https://proceedings.mlr.press/v235/ruaud24a.html
|
ICML 2024
|
Understanding the interactions and interplay of microorganisms is a great challenge with many applications in medical and environmental settings. In this work, we model bacterial communities directly from their genomes using graph neural networks (GNNs). GNNs leverage the inductive bias induced by the set nature of bacteria, enforcing permutation invariance and granting combinatorial generalization. We propose to learn the dynamics implicitly by directly predicting community relative abundance profiles at steady state, thus escaping the need for growth curves. On two real-world datasets, we show for the first time generalization to unseen bacteria and different community structures. To investigate the prediction results more deeply, we create a simulation for flexible data generation and analyze effects of bacteria interaction strength, community size, and training data amount.
|
https://proceedings.mlr.press/v235/rudikov24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rudikov24a/rudikov24a.pdf
|
https://openreview.net/forum?id=J0ty1o7nCj
|
Neural operators meet conjugate gradients: The FCG-NO method for efficient PDE solving
|
https://proceedings.mlr.press/v235/rudikov24a.html
|
Alexander Rudikov, Vladimir Fanaskov, Ekaterina Muravleva, Yuri M. Laevsky, Ivan Oseledets
|
https://proceedings.mlr.press/v235/rudikov24a.html
|
ICML 2024
|
Deep learning solvers for partial differential equations typically have limited accuracy. We propose to overcome this problem by using them as preconditioners. More specifically, we apply discretization-invariant neural operators to learn preconditioners for the flexible conjugate gradient method (FCG). Architecture paired with novel loss function and training scheme allows for learning efficient preconditioners that can be used across different resolutions. On the theoretical side, FCG theory allows us to safely use nonlinear preconditioners that can be applied in $O(N)$ operations without constraining the form of the preconditioners matrix. To justify learning scheme components (the loss function and the way training data is collected) we perform several ablation studies. Numerical results indicate that our approach favorably compares with classical preconditioners and allows to reuse of preconditioners learned for lower resolution to the higher resolution data.
|
https://proceedings.mlr.press/v235/rudin24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rudin24a/rudin24a.pdf
|
https://openreview.net/forum?id=oFDFGd9Age
|
Position: Amazing Things Come From Having Many Good Models
|
https://proceedings.mlr.press/v235/rudin24a.html
|
Cynthia Rudin, Chudi Zhong, Lesia Semenova, Margo Seltzer, Ronald Parr, Jiachang Liu, Srikar Katta, Jon Donnelly, Harry Chen, Zachery Boner
|
https://proceedings.mlr.press/v235/rudin24a.html
|
ICML 2024
|
The Rashomon Effect, coined by Leo Breiman, describes the phenomenon that there exist many equally good predictive models for the same dataset. This phenomenon happens for many real datasets and when it does, it sparks both magic and consternation, but mostly magic. In light of the Rashomon Effect, this perspective piece proposes reshaping the way we think about machine learning, particularly for tabular data problems in the nondeterministic (noisy) setting. We address how the Rashomon Effect impacts (1) the existence of simple-yet-accurate models, (2) flexibility to address user preferences, such as fairness and monotonicity, without losing performance, (3) uncertainty in predictions, fairness, and explanations, (4) reliable variable importance, (5) algorithm choice, specifically, providing advanced knowledge of which algorithms might be suitable for a given problem, and (6) public policy. We also discuss a theory of when the Rashomon Effect occurs and why. Our goal is to illustrate how the Rashomon Effect can have a massive impact on the use of machine learning for complex problems in society.
|
https://proceedings.mlr.press/v235/rugamer24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rugamer24a/rugamer24a.pdf
|
https://openreview.net/forum?id=p9SMltcfsu
|
Generalizing Orthogonalization for Models with Non-Linearities
|
https://proceedings.mlr.press/v235/rugamer24a.html
|
David Rügamer, Chris Kolb, Tobias Weber, Lucas Kook, Thomas Nagler
|
https://proceedings.mlr.press/v235/rugamer24a.html
|
ICML 2024
|
The complexity of black-box algorithms can lead to various challenges, including the introduction of biases. These biases present immediate risks in the algorithms’ application. It was, for instance, shown that neural networks can deduce racial information solely from a patient’s X-ray scan, a task beyond the capability of medical experts. If this fact is not known to the medical expert, automatic decision-making based on this algorithm could lead to prescribing a treatment (purely) based on racial information. While current methodologies allow for the "orthogonalization" or "normalization" of neural networks with respect to such information, existing approaches are grounded in linear models. Our paper advances the discourse by introducing corrections for non-linearities such as ReLU activations. Our approach also encompasses scalar and tensor-valued predictions, facilitating its integration into neural network architectures. Through extensive experiments, we validate our method’s effectiveness in safeguarding sensitive data in generalized linear models, normalizing convolutional neural networks for metadata, and rectifying pre-existing embeddings for undesired attributes.
|
https://proceedings.mlr.press/v235/ruhe24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/ruhe24a/ruhe24a.pdf
|
https://openreview.net/forum?id=a9bzTv9SzO
|
Rolling Diffusion Models
|
https://proceedings.mlr.press/v235/ruhe24a.html
|
David Ruhe, Jonathan Heek, Tim Salimans, Emiel Hoogeboom
|
https://proceedings.mlr.press/v235/ruhe24a.html
|
ICML 2024
|
Diffusion models have recently been increasingly applied to temporal data such as video, fluid mechanics simulations, or climate data. These methods generally treat subsequent frames equally regarding the amount of noise in the diffusion process. This paper explores Rolling Diffusion: a new approach that uses a sliding window denoising process. It ensures that the diffusion process progressively corrupts through time by assigning more noise to frames that appear later in a sequence, reflecting greater uncertainty about the future as the generation process unfolds. Empirically, we show that when the temporal dynamics are complex, Rolling Diffusion is superior to standard diffusion. In particular, this result is demonstrated in a video prediction task using the Kinetics-600 video dataset and in a chaotic fluid dynamics forecasting experiment.
|
https://proceedings.mlr.press/v235/rushing24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/rushing24a/rushing24a.pdf
|
https://openreview.net/forum?id=5ZwEifshyo
|
Explorations of Self-Repair in Language Models
|
https://proceedings.mlr.press/v235/rushing24a.html
|
Cody Rushing, Neel Nanda
|
https://proceedings.mlr.press/v235/rushing24a.html
|
ICML 2024
|
Prior interpretability research studying narrow distributions has preliminarily identified self-repair, a phenomena where if components in large language models are ablated, later components will change their behavior to compensate. Our work builds off this past literature, demonstrating that self-repair exists on a variety of models families and sizes when ablating individual attention heads on the full training distribution. We further show that on the full training distribution self-repair is imperfect, as the original direct effect of the head is not fully restored, and noisy, since the degree of self-repair varies significantly across different prompts (sometimes overcorrecting beyond the original effect). We highlight two different mechanisms that contribute to self-repair, including changes in the final LayerNorm scaling factor and sparse sets of neurons implementing Anti-Erasure. We additionally discuss the implications of these results for interpretability practitioners and close with a more speculative discussion on the mystery of why self-repair occurs in these models at all, highlighting evidence for the Iterative Inference hypothesis in language models, a framework that predicts self-repair.
|
https://proceedings.mlr.press/v235/ryu24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/ryu24a/ryu24a.pdf
|
https://openreview.net/forum?id=mu7Er7f9NQ
|
Gambling-Based Confidence Sequences for Bounded Random Vectors
|
https://proceedings.mlr.press/v235/ryu24a.html
|
Jongha Jon Ryu, Gregory W. Wornell
|
https://proceedings.mlr.press/v235/ryu24a.html
|
ICML 2024
|
A confidence sequence (CS) is a sequence of confidence sets that contains a target parameter of an underlying stochastic process at any time step with high probability. This paper proposes a new approach to constructing CSs for means of bounded multivariate stochastic processes using a general gambling framework, extending the recently established coin toss framework for bounded random processes. The proposed gambling framework provides a general recipe for constructing CSs for categorical and probability-vector-valued observations, as well as for general bounded multidimensional observations through a simple reduction. This paper specifically explores the use of the mixture portfolio, akin to Cover’s universal portfolio, in the proposed framework and investigates the properties of the resulting CSs. Simulations demonstrate the tightness of these confidence sequences compared to existing methods. When applied to the sampling without-replacement setting for finite categorical data, it is shown that the resulting CS based on a universal gambling strategy is provably tighter than that of the posterior-prior ratio martingale proposed by Waudby-Smith and Ramdas.
|
https://proceedings.mlr.press/v235/ryu24b.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/ryu24b/ryu24b.pdf
|
https://openreview.net/forum?id=qESG5HaaoJ
|
Operator SVD with Neural Networks via Nested Low-Rank Approximation
|
https://proceedings.mlr.press/v235/ryu24b.html
|
Jongha Jon Ryu, Xiangxiang Xu, Hasan Sabri Melihcan Erol, Yuheng Bu, Lizhong Zheng, Gregory W. Wornell
|
https://proceedings.mlr.press/v235/ryu24b.html
|
ICML 2024
|
Computing eigenvalue decomposition (EVD) of a given linear operator, or finding its leading eigenvalues and eigenfunctions, is a fundamental task in many machine learning and scientific simulation problems. For high-dimensional eigenvalue problems, training neural networks to parameterize the eigenfunctions is considered as a promising alternative to the classical numerical linear algebra techniques. This paper proposes a new optimization framework based on the low-rank approximation characterization of a truncated singular value decomposition, accompanied by new techniques called nesting for learning the top-$L$ singular values and singular functions in the correct order. The proposed method promotes the desired orthogonality in the learned functions implicitly and efficiently via an unconstrained optimization formulation, which is easy to solve with off-the-shelf gradient-based optimization algorithms. We demonstrate the effectiveness of the proposed optimization framework for use cases in computational physics and machine learning.
|
https://proceedings.mlr.press/v235/s24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/s24a/s24a.pdf
|
https://openreview.net/forum?id=TN3fi7dwPo
|
Tandem Transformers for Inference Efficient LLMs
|
https://proceedings.mlr.press/v235/s24a.html
|
Aishwarya P S, Pranav Ajit Nair, Yashas Samaga B L, Toby James Boyd, Sanjiv Kumar, Prateek Jain, Praneeth Netrapalli
|
https://proceedings.mlr.press/v235/s24a.html
|
ICML 2024
|
The autoregressive nature of conventional large language models (LLMs) inherently limits inference speed, as tokens are generated sequentially. While speculative (Leviathan et al., 2023) and parallel (Stern et al., 2018) decoding techniques attempt to mitigate this, they face limitations: either relying on less accurate smaller models for generation or failing to fully leverage the base LLM’s representations. We introduce a novel architecture, Tandem transformers, to address these issues. This architecture uniquely combines (1) a small autoregressive model and (2) a large model operating in block mode (processing multiple tokens simultaneously). The small model’s predictive accuracy is substantially enhanced by granting it attention to the large model’s richer representations. On the PaLM2 pretraining dataset, a tandem of PaLM2-Bison and PaLM2-Gecko demonstrates a 3.3% improvement in next-token prediction accuracy over a standalone PaLM2-Gecko, offering a 1.16x speedup compared to a PaLM2-Otter model with comparable downstream performance. We further incorporate the Tandem model within the speculative decoding (SPEED) framework where the large model validates tokens from the small model. This ensures that the tandem of PaLM2-Bison and PaLM2-Gecko achieves substantial speedup (around 1.14x faster than using vanilla PaLM2-Gecko in SPEED) while maintaining identical downstream task accuracy.
|
https://proceedings.mlr.press/v235/saad-falcon24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/saad-falcon24a/saad-falcon24a.pdf
|
https://openreview.net/forum?id=HkCRgoGtt6
|
Benchmarking and Building Long-Context Retrieval Models with LoCo and M2-BERT
|
https://proceedings.mlr.press/v235/saad-falcon24a.html
|
Jon Saad-Falcon, Daniel Y Fu, Simran Arora, Neel Guha, Christopher Re
|
https://proceedings.mlr.press/v235/saad-falcon24a.html
|
ICML 2024
|
Retrieval pipelines are an integral component of many machine learning systems. However, they perform poorly in domains where documents are long (e.g., 10K tokens or more) and where identifying the relevant document requires synthesizing information across the entire text. Developing long-context retrieval encoders suitable for these domains raises three challenges: (1) how to evaluate long-context retrieval performance, (2) how to pretrain a base language model to represent both short contexts (corresponding to queries) and long contexts (corresponding to documents), and (3) how to finetune this model for retrieval under the batch size limitations imposed by GPU memory constraints. To address these challenges, we first introduce LoCoV1, a 12 task benchmark constructed to measure long-context retrieval where chunking is not possible or not effective. We next present the M2-BERT retrieval encoder, an 80M parameter state-space encoder model built from the Monarch Mixer architecture, capable of scaling to documents up to 32K tokens long. We describe a pretraining data mixture which allows this encoder to process both short and long context sequences, and a finetuning approach that adapts this base model to retrieval with only single-sample batches. Finally, we validate the M2-BERT retrieval encoder on LoCoV1, finding that it outperforms competitive Transformer-based models by at least 22.2 points, despite containing 90× fewer parameters.
|
https://proceedings.mlr.press/v235/sabour24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sabour24a/sabour24a.pdf
|
https://openreview.net/forum?id=nBGBzV4It3
|
Align Your Steps: Optimizing Sampling Schedules in Diffusion Models
|
https://proceedings.mlr.press/v235/sabour24a.html
|
Amirmojtaba Sabour, Sanja Fidler, Karsten Kreis
|
https://proceedings.mlr.press/v235/sabour24a.html
|
ICML 2024
|
Diffusion models (DMs) have established themselves as the state-of-the-art generative modeling approach in the visual domain and beyond. A crucial drawback of DMs is their slow sampling speed, relying on many sequential function evaluations through large neural networks. Sampling from DMs can be seen as solving a differential equation through a discretized set of noise levels known as the sampling schedule. While past works primarily focused on deriving efficient solvers, little attention has been given to finding optimal sampling schedules, and the entire literature relies on hand-crafted heuristics. In this work, for the first time, we propose a general and principled approach to optimizing the sampling schedules of DMs for high-quality outputs, called Align Your Steps. We leverage methods from stochastic calculus and find optimal schedules specific to different solvers, trained DMs and datasets. We evaluate our novel approach on several image, video as well as 2D toy data synthesis benchmarks, using a variety of different samplers, and observe that our optimized schedules outperform previous hand-crafted schedules in almost all experiments. Our method demonstrates the untapped potential of sampling schedule optimization, especially in the few-step synthesis regime.
|
https://proceedings.mlr.press/v235/sadasivan24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sadasivan24a/sadasivan24a.pdf
|
https://openreview.net/forum?id=wCMNbdshcY
|
Fast Adversarial Attacks on Language Models In One GPU Minute
|
https://proceedings.mlr.press/v235/sadasivan24a.html
|
Vinu Sankar Sadasivan, Shoumik Saha, Gaurang Sriramanan, Priyatham Kattakinda, Atoosa Chegini, Soheil Feizi
|
https://proceedings.mlr.press/v235/sadasivan24a.html
|
ICML 2024
|
In this paper, we introduce a novel class of fast, beam search-based adversarial attack (BEAST) for Language Models (LMs). BEAST employs interpretable parameters, enabling attackers to balance between attack speed, success rate, and the readability of adversarial prompts. The computational efficiency of BEAST facilitates us to investigate its applications on LMs for jailbreaking, eliciting hallucinations, and privacy attacks. Our gradient-free targeted attack can jailbreak aligned LMs with high attack success rates within one minute. For instance, BEAST can jailbreak Vicuna-7B-v1.5 under one minute with a success rate of 89% when compared to a gradient-based baseline that takes over an hour to achieve 70% success rate using a single Nvidia RTX A6000 48GB GPU. BEAST can also generate adversarial suffixes for successful jailbreaks that can transfer to unseen prompts and unseen models such as GPT-4-Turbo. Additionally, we discover a unique outcome wherein our untargeted attack induces hallucinations in LM chatbots. Through human evaluations, we find that our untargeted attack causes Vicuna-7B-v1.5 to produce $\sim$15% more incorrect outputs when compared to LM outputs in the absence of our attack. We also learn that 22% of the time, BEAST causes Vicuna to generate outputs that are not relevant to the original prompt. Further, we use BEAST to generate adversarial prompts in a few seconds that can boost the performance of existing membership inference attacks for LMs. We believe that our fast attack, BEAST, has the potential to accelerate research in LM security and privacy.
|
https://proceedings.mlr.press/v235/sagar24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sagar24a/sagar24a.pdf
|
https://openreview.net/forum?id=DkqiId4AuR
|
Failures Are Fated, But Can Be Faded: Characterizing and Mitigating Unwanted Behaviors in Large-Scale Vision and Language Models
|
https://proceedings.mlr.press/v235/sagar24a.html
|
Som Sagar, Aditya Taparia, Ransalu Senanayake
|
https://proceedings.mlr.press/v235/sagar24a.html
|
ICML 2024
|
In large deep neural networks that seem to perform surprisingly well on many tasks, we also observe a few failures related to accuracy, social biases, and alignment with human values, among others. Therefore, before deploying these models, it is crucial to characterize this failure landscape for engineers to debug and legislative bodies to audit models. Nevertheless, it is infeasible to exhaustively test for all possible combinations of factors that could lead to a model’s failure. In this paper, we introduce a post-hoc method that utilizes deep reinforcement learning to explore and construct the landscape of failure modes in pre-trained discriminative and generative models. With the aid of limited human feedback, we then demonstrate how to restructure the failure landscape to be more desirable by moving away from the discovered failure modes. We empirically show the effectiveness of the proposed method across common Computer Vision, Natural Language Processing, and Vision-Language tasks.
|
https://proceedings.mlr.press/v235/saha24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/saha24a/saha24a.pdf
|
https://openreview.net/forum?id=MdPBVWTfwG
|
I/O Complexity of Attention, or How Optimal is FlashAttention?
|
https://proceedings.mlr.press/v235/saha24a.html
|
Barna Saha, Christopher Ye
|
https://proceedings.mlr.press/v235/saha24a.html
|
ICML 2024
|
Attention is at the heart of the popular Transformer architecture, yet suffers from quadratic time and memory complexity. In a recent significant development, FlashAttention shows that the I/O complexity of attention is the true bottleneck in scaling Transformers. Given two levels of memory hierarchy, a fast cache (e.g. GPU on-chip SRAM) where computation happens and a slow memory (e.g. GPU high-bandwidth memory) where the data resides, the I/O complexity measures the number of accesses to the slow memory. FlashAttention is an I/O-aware algorithm for self-attention that requires $\frac{N^2d^2}{M}$ I/O operations where $N$ is the dimension of the attention matrix, $d$ is the head-dimension and $M$ is the size of cache. Naturally, to further reduce the computational costs of Attention, the authors ask the question: is FlashAttention’s I/O complexity optimal for every value of $M$? We resolve the above question in its full generality by showing an I/O complexity lower bound that matches the upper bound provided by FlashAttention for any values of $M \geq d^2$ within any constant factors. Moreover, our lower bounds do not rely on using combinatorial matrix multiplication for computing the attention matrix: even if one uses fast matrix multiplication, the above I/O complexity bounds cannot be improved. Further, we give a better algorithm with lower I/O complexity for $M < d^2$, and show that it is optimal for combinatorial algorithms. We do so by introducing a new communication complexity protocol for matrix compression, and connecting communication complexity to I/O complexity. We believe this connection could be of independent interest and will find more applications in proving I/O complexity lower bounds in future.
|
https://proceedings.mlr.press/v235/sahinoglu24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sahinoglu24a/sahinoglu24a.pdf
|
https://openreview.net/forum?id=PSzyBN7LIA
|
An Online Optimization Perspective on First-Order and Zero-Order Decentralized Nonsmooth Nonconvex Stochastic Optimization
|
https://proceedings.mlr.press/v235/sahinoglu24a.html
|
Emre Sahinoglu, Shahin Shahrampour
|
https://proceedings.mlr.press/v235/sahinoglu24a.html
|
ICML 2024
|
We investigate the finite-time analysis of finding ($\delta, \epsilon$)-stationary points for nonsmooth nonconvex objectives in decentralized stochastic optimization. A set of agents aim at minimizing a global function using only their local information by interacting over a network. We present a novel algorithm, called Multi Epoch Decentralized Online Learning (ME-DOL), for which we establish the sample complexity in various settings. First, using a recently proposed online-to-nonconvex technique, we show that our algorithm recovers the optimal convergence rate of smooth nonconvex objectives. We then extend our analysis to the nonsmooth setting, building on properties of randomized smoothing and Goldstein-subdifferential sets. We establish the sample complexity of $O(\delta^{-1}\epsilon^{-3})$, which to the best of our knowledge is the first finite-time guarantee for decentralized nonsmooth nonconvex stochastic optimization in the first-order setting (without weak-convexity), matching its optimal centralized counterpart. We further prove the same rate for the zero-order oracle setting without using variance reduction.
|
https://proceedings.mlr.press/v235/sale24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sale24a/sale24a.pdf
|
https://openreview.net/forum?id=VJjjNrUi8j
|
Second-Order Uncertainty Quantification: A Distance-Based Approach
|
https://proceedings.mlr.press/v235/sale24a.html
|
Yusuf Sale, Viktor Bengs, Michele Caprio, Eyke Hüllermeier
|
https://proceedings.mlr.press/v235/sale24a.html
|
ICML 2024
|
In the past couple of years, various approaches to representing and quantifying different types of predictive uncertainty in machine learning, notably in the setting of classification, have been proposed on the basis of second-order probability distributions, i.e., predictions in the form of distributions on probability distributions. A completely conclusive solution has not yet been found, however, as shown by recent criticisms of commonly used uncertainty measures associated with second-order distributions, identifying undesirable theoretical properties of these measures. In light of these criticisms, we propose a set of formal criteria that meaningful uncertainty measures for predictive uncertainty based on second-order distributions should obey. Moreover, we provide a general framework for developing uncertainty measures to account for these criteria, and offer an instantiation based on the Wasserstein distance, for which we prove that all criteria are satisfied.
|
https://proceedings.mlr.press/v235/saleh24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/saleh24a/saleh24a.pdf
|
https://openreview.net/forum?id=itDhUBY2xf
|
Learning from Integral Losses in Physics Informed Neural Networks
|
https://proceedings.mlr.press/v235/saleh24a.html
|
Ehsan Saleh, Saba Ghaffari, Tim Bretl, Luke Olson, Matthew West
|
https://proceedings.mlr.press/v235/saleh24a.html
|
ICML 2024
|
This work proposes a solution for the problem of training physics-informed networks under partial integro-differential equations. These equations require an infinite or a large number of neural evaluations to construct a single residual for training. As a result, accurate evaluation may be impractical, and we show that naive approximations at replacing these integrals with unbiased estimates lead to biased loss functions and solutions. To overcome this bias, we investigate three types of potential solutions: the deterministic sampling approaches, the double-sampling trick, and the delayed target method. We consider three classes of PDEs for benchmarking; one defining Poisson problems with singular charges and weak solutions of up to 10 dimensions, another involving weak solutions on electro-magnetic fields and a Maxwell equation, and a third one defining a Smoluchowski coagulation problem. Our numerical results confirm the existence of the aforementioned bias in practice and also show that our proposed delayed target approach can lead to accurate solutions with comparable quality to ones estimated with a large sample size integral. Our implementation is open-source and available at https://github.com/ehsansaleh/btspinn.
|
https://proceedings.mlr.press/v235/salgia24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/salgia24a/salgia24a.pdf
|
https://openreview.net/forum?id=tOO6PD3kYP
|
Random Exploration in Bayesian Optimization: Order-Optimal Regret and Computational Efficiency
|
https://proceedings.mlr.press/v235/salgia24a.html
|
Sudeep Salgia, Sattar Vakili, Qing Zhao
|
https://proceedings.mlr.press/v235/salgia24a.html
|
ICML 2024
|
We consider Bayesian optimization using Gaussian Process models, also referred to as kernel-based bandit optimization. We study the methodology of exploring the domain using random samples drawn from a distribution. We show that this random exploration approach achieves the optimal error rates. Our analysis is based on novel concentration bounds in an infinite dimensional Hilbert space established in this work, which may be of independent interest. We further develop an algorithm based on random exploration with domain shrinking and establish its order-optimal regret guarantees under both noise-free and noisy settings. In the noise-free setting, our analysis closes the existing gap in regret performance under a mild assumption on the underlying function and thereby partially resolves a COLT open problem. The proposed algorithm also enjoys a computational advantage over prevailing methods due to the random exploration that obviates the expensive optimization of a non-convex acquisition function for choosing the query points at each iteration.
|
https://proceedings.mlr.press/v235/salvatori24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/salvatori24a/salvatori24a.pdf
|
https://openreview.net/forum?id=nTgzmXvuEA
|
Predictive Coding beyond Correlations
|
https://proceedings.mlr.press/v235/salvatori24a.html
|
Tommaso Salvatori, Luca Pinchetti, Amine M’Charrak, Beren Millidge, Thomas Lukasiewicz
|
https://proceedings.mlr.press/v235/salvatori24a.html
|
ICML 2024
|
Biologically plausible learning algorithms offer a promising alternative to traditional deep learning techniques, especially in overcoming the limitations of backpropagation in fast and low-energy neuromorphic implementations. To this end, there has been extensive research in understanding what their capabilities are. In this work, we show how one of such algorithms, called predictive coding, is able to perform causal inference tasks. First, we show how a simple change in the inference process of predictive coding enables to compute interventions without the need to mutilate or redefine a causal graph. Then, we explore applications in cases where the graph is unknown, and has to be inferred from observational data. Empirically, we show how such findings can be used to improve the performance of predictive coding in image classification tasks, and conclude that such models are naturally able to perform causal inference tasks using a biologically plausible kind of message passing.
|
https://proceedings.mlr.press/v235/san-roman24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/san-roman24a/san-roman24a.pdf
|
https://openreview.net/forum?id=Bic3Vmy2DG
|
Proactive Detection of Voice Cloning with Localized Watermarking
|
https://proceedings.mlr.press/v235/san-roman24a.html
|
Robin San Roman, Pierre Fernandez, Hady Elsahar, Alexandre Défossez, Teddy Furon, Tuan Tran
|
https://proceedings.mlr.press/v235/san-roman24a.html
|
ICML 2024
|
In the rapidly evolving field of speech generative models, there is a pressing need to ensure audio authenticity against the risks of voice cloning. We present AudioSeal, the first audio watermarking technique designed specifically for localized detection of AI-generated speech. AudioSeal employs a generator / detector architecture trained jointly with a localization loss to enable localized watermark detection up to the sample level, and a novel perceptual loss inspired by auditory masking, that enables AudioSeal to achieve better imperceptibility. AudioSeal achieves state-of-the-art performance in terms of robustness to real life audio manipulations and imperceptibility based on automatic and human evaluation metrics. Additionally, AudioSeal is designed with a fast, single-pass detector, that significantly surpasses existing models in speed, achieving detection up to two orders of magnitude faster, making it ideal for large-scale and real-time applications.Code is available at https://github.com/facebookresearch/audioseal
|
https://proceedings.mlr.press/v235/sanchez24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sanchez24a/sanchez24a.pdf
|
https://openreview.net/forum?id=RiM3cl9MdK
|
Stay on Topic with Classifier-Free Guidance
|
https://proceedings.mlr.press/v235/sanchez24a.html
|
Guillaume Sanchez, Alexander Spangher, Honglu Fan, Elad Levi, Stella Biderman
|
https://proceedings.mlr.press/v235/sanchez24a.html
|
ICML 2024
|
Classifier-Free Guidance (CFG) has recently emerged in as a lightweight technique to encourage prompt-adherence in generations, yet has not yet been successfully applied to language modeling. In this work, we demonstrate across a wide array of benchmarks that CFG can be used broadly as an inference-time technique in pure language modeling. We show that CFG (1) improves the performance of Pythia, GPT-2 and LLaMA-family models across: Q&A, reasoning, code generation, and machine translation, achieving SOTA on LAMBADA with LLaMA-7B over PaLM-540B; (2) brings improvements equivalent to a model with twice the parameter-count; (3) can stack alongside other inference-time methods like Chain-of-Thought and Self-Consistency, yielding further improvements in difficult tasks; (4) can be used to increase the faithfulness and coherence of assistants in challenging form-driven and content-driven prompts: in human evaluations we show a 75% preference for using CFG over baseline.
|
https://proceedings.mlr.press/v235/sander24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sander24a/sander24a.pdf
|
https://openreview.net/forum?id=kZbTkpnafR
|
How do Transformers Perform In-Context Autoregressive Learning ?
|
https://proceedings.mlr.press/v235/sander24a.html
|
Michael Eli Sander, Raja Giryes, Taiji Suzuki, Mathieu Blondel, Gabriel Peyré
|
https://proceedings.mlr.press/v235/sander24a.html
|
ICML 2024
|
Transformers have achieved state-of-the-art performance in language modeling tasks. However, the reasons behind their tremendous success are still unclear. In this paper, towards a better understanding, we train a Transformer model on a simple next token prediction task, where sequences are generated as a first-order autoregressive process $s_{t+1} = W s_t$. We show how a trained Transformer predicts the next token by first learning $W$ in-context, then applying a prediction mapping. We call the resulting procedure in-context autoregressive learning. More precisely, focusing on commuting orthogonal matrices $W$, we first show that a trained one-layer linear Transformer implements one step of gradient descent for the minimization of an inner objective function, when considering augmented tokens. When the tokens are not augmented, we characterize the global minima of a one-layer diagonal linear multi-head Transformer. Importantly, we exhibit orthogonality between heads and show that positional encoding captures trigonometric relations in the data. On the experimental side, we consider the general case of non-commuting orthogonal matrices and generalize our theoretical findings.
|
https://proceedings.mlr.press/v235/sander24b.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sander24b/sander24b.pdf
|
https://openreview.net/forum?id=Nw7yOe8nBi
|
Differentially Private Representation Learning via Image Captioning
|
https://proceedings.mlr.press/v235/sander24b.html
|
Tom Sander, Yaodong Yu, Maziar Sanjabi, Alain Oliviero Durmus, Yi Ma, Kamalika Chaudhuri, Chuan Guo
|
https://proceedings.mlr.press/v235/sander24b.html
|
ICML 2024
|
Differentially private (DP) machine learning is considered the gold-standard solution for training a model from sensitive data while still preserving privacy. However, a major barrier to achieving this ideal is its sub-optimal privacy-accuracy trade-off, which is particularly visible in DP representation learning. Specifically, it has been shown that under modest privacy budgets, most models learn representations that are not significantly better than hand-crafted features. In this work, we show that effective DP representation learning can be done via image captioning and scaling up to internet-scale multimodal datasets. Through a series of engineering tricks, we successfully train a DP image captioner (DP-Cap) on a 233M subset of LAION-2B from scratch using a reasonable amount of computation, and obtaining unprecedented high-quality image features that can be used in a variety of downstream vision and vision-language tasks. For example, under a privacy budget of $\varepsilon=8$ for the LAION dataset, a linear classifier trained on top of learned DP-Cap features attains $65.8%$ accuracy on ImageNet-1K, considerably improving the previous SOTA of $56.5%$. Our work challenges the prevailing sentiment that high-utility DP representation learning cannot be achieved by training from scratch.
|
https://proceedings.mlr.press/v235/sanford24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sanford24a/sanford24a.pdf
|
https://openreview.net/forum?id=QCZabhKQhB
|
Transformers, parallel computation, and logarithmic depth
|
https://proceedings.mlr.press/v235/sanford24a.html
|
Clayton Sanford, Daniel Hsu, Matus Telgarsky
|
https://proceedings.mlr.press/v235/sanford24a.html
|
ICML 2024
|
We show that a constant number of self-attention layers can efficiently simulate—and be simulated by—a constant number of communication rounds of Massively Parallel Computation. As a consequence, we show that logarithmic-depth is sufficient for transformers to solve basic computational tasks that cannot be efficiently solved by several other neural sequence models and sub-quadratic transformer approximations. We thus establish parallelism as a key distinguishing property of transformers.
|
https://proceedings.mlr.press/v235/sankararaman24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sankararaman24a/sankararaman24a.pdf
|
https://openreview.net/forum?id=1OsRSrkFWl
|
Promoting External and Internal Equities Under Ex-Ante/Ex-Post Metrics in Online Resource Allocation
|
https://proceedings.mlr.press/v235/sankararaman24a.html
|
Karthik Abinav Sankararaman, Aravind Srinivasan, Pan Xu
|
https://proceedings.mlr.press/v235/sankararaman24a.html
|
ICML 2024
|
This paper proposes two different models for equitable resource allocation in online settings. The first one is called external equity promotion, where sequentially arriving agents are heterogeneous in their external attributes, namely how many resources they demand, which are drawn from a probability distribution (accessible to the algorithm). The focus is then to devise an allocation policy such that every requester can get a fair share of resources proportional to their demands, regardless of their arrival time. The second is called internal equity promotion, where arriving requesters can be treated homogeneously in external attributes (demands) but are heterogeneous in internal traits such as demographics. In particular, each requester can be identified as belonging to one or several groups, and an allocation of resources is regarded as equitable when every group of requesters can receive a fair share of resources proportional to the percentage of that group in the whole population. For both models above, we consider as the benchmark a clairvoyant optimal solution that has the privilege to access all random demand realizations in advance. We consider two equity metrics, namely ex-post and ex-ante, and discuss the challenges under the two metrics in detail. Specifically, we present two linear program (LP)-based policies for external equity promotion under ex-ante with independent demands, each achieving an optimal CR of $1/2$ with respect to the benchmark LP. For internal equity promotion, we present optimal policies under both ex-ante and ex-post metrics.
|
https://proceedings.mlr.press/v235/sanokowski24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sanokowski24a/sanokowski24a.pdf
|
https://openreview.net/forum?id=AFfXlKFHXJ
|
A Diffusion Model Framework for Unsupervised Neural Combinatorial Optimization
|
https://proceedings.mlr.press/v235/sanokowski24a.html
|
Sebastian Sanokowski, Sepp Hochreiter, Sebastian Lehner
|
https://proceedings.mlr.press/v235/sanokowski24a.html
|
ICML 2024
|
Learning to sample from intractable distributions over discrete sets without relying on corresponding training data is a central problem in a wide range of fields, including Combinatorial Optimization. Currently, popular deep learning-based approaches rely primarily on generative models that yield exact sample likelihoods. This work introduces a method that lifts this restriction and opens the possibility to employ highly expressive latent variable models like diffusion models. Our approach is conceptually based on a loss that upper bounds the reverse Kullback-Leibler divergence and evades the requirement of exact sample likelihoods. We experimentally validate our approach in data-free Combinatorial Optimization and demonstrate that our method achieves a new state-of-the-art on a wide range of benchmark problems.
|
https://proceedings.mlr.press/v235/santos24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/santos24a/santos24a.pdf
|
https://openreview.net/forum?id=OdPlFWExX1
|
Sparse and Structured Hopfield Networks
|
https://proceedings.mlr.press/v235/santos24a.html
|
Saul José Rodrigues Dos Santos, Vlad Niculae, Daniel C Mcnamee, Andre Martins
|
https://proceedings.mlr.press/v235/santos24a.html
|
ICML 2024
|
Modern Hopfield networks have enjoyed recent interest due to their connection to attention in transformers. Our paper provides a unified framework for sparse Hopfield networks by establishing a link with Fenchel-Young losses. The result is a new family of Hopfield-Fenchel-Young energies whose update rules are end-to-end differentiable sparse transformations. We reveal a connection between loss margins, sparsity, and exact memory retrieval. We further extend this framework to structured Hopfield networks via the SparseMAP transformation, which can retrieve pattern associations instead of a single pattern. Experiments on multiple instance learning and text rationalization demonstrate the usefulness of our approach.
|
https://proceedings.mlr.press/v235/sapkota24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sapkota24a/sapkota24a.pdf
|
https://openreview.net/forum?id=CquFGSIU6w
|
Meta Evidential Transformer for Few-Shot Open-Set Recognition
|
https://proceedings.mlr.press/v235/sapkota24a.html
|
Hitesh Sapkota, Krishna Prasad Neupane, Qi Yu
|
https://proceedings.mlr.press/v235/sapkota24a.html
|
ICML 2024
|
Few-shot open-set recognition (FSOSR) aims to detect instances from unseen classes by utilizing a small set of labeled instances from closed-set classes. Accurately rejecting instances from open-set classes in the few-shot setting is fundamentally more challenging due to the weaker supervised signals resulting from fewer labels. Transformer-based few-shot methods exploit attention mapping to achieve a consistent representation. However, the softmax-generated attention map normalizes all the instances that assign unnecessary high attentive weights to those instances not close to the closed-set classes that negatively impact the detection performance. In addition, open-set samples that are similar to a certain closed-set class also pose a significant challenge to most existing FSOSR models. To address these challenges, we propose a novel Meta Evidential Transformer (MET) based FSOSR model that uses an evidential open-set loss to learn more compact closed-set class representations by effectively leveraging similar closed-set classes. MET further integrates an evidence-to-variance ratio to detect fundamentally challenging tasks and uses an evidence-guided cross-attention mechanism to better separate the difficult open-set samples. Experiments on real-world datasets demonstrate consistent improvement over existing competitive methods in unseen class recognition without deteriorating closed-set performance.
|
https://proceedings.mlr.press/v235/sapora24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sapora24a/sapora24a.pdf
|
https://openreview.net/forum?id=9DMMvMTDur
|
EvIL: Evolution Strategies for Generalisable Imitation Learning
|
https://proceedings.mlr.press/v235/sapora24a.html
|
Silvia Sapora, Gokul Swamy, Chris Lu, Yee Whye Teh, Jakob Nicolaus Foerster
|
https://proceedings.mlr.press/v235/sapora24a.html
|
ICML 2024
|
Often times in imitation learning (IL), the environment we collect expert demonstrations in and the environment we want to deploy our learned policy in aren’t exactly the same (e.g. demonstrations collected in simulation but deployment in the real world). Compared to policy-centric approaches to IL like behavioural cloning, reward-centric approaches like inverse reinforcement learning (IRL) often better replicate expert behaviour in new environments. This transfer is usually performed by optimising the recovered reward under the dynamics of the target environment. However, (a) we find that modern deep IL algorithms frequently recover rewards which induce policies far weaker than the expert, even in the same environment the demonstrations were collected in. Furthermore, (b) these rewards are often quite poorly shaped, necessitating extensive environment interaction to optimise effectively. We provide simple and scalable fixes to both of these concerns. For (a), we find that reward model ensembles combined with a slightly different training objective significantly improves re-training and transfer performance. For (b), we propose a novel evolution-strategies based method (EvIL) to optimise for a reward-shaping term that speeds up re-training in the target environment, closing a gap left open by the classical theory of IRL. On a suite of continuous control tasks, we are able to re-train policies in target (and source) environments more interaction-efficiently than prior work.
|
https://proceedings.mlr.press/v235/saratchandran24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/saratchandran24a/saratchandran24a.pdf
|
https://openreview.net/forum?id=cVkqItmYLQ
|
A sampling theory perspective on activations for implicit neural representations
|
https://proceedings.mlr.press/v235/saratchandran24a.html
|
Hemanth Saratchandran, Sameera Ramasinghe, Violetta Shevchenko, Alexander Long, Simon Lucey
|
https://proceedings.mlr.press/v235/saratchandran24a.html
|
ICML 2024
|
Implicit Neural Representations (INRs) have gained popularity for encoding signals as compact, differentiable entities. While commonly using techniques like Fourier positional encodings or non-traditional activation functions (e.g., Gaussian, sinusoid, or wavelets) to capture high-frequency content, their properties lack exploration within a unified theoretical framework. Addressing this gap, we conduct a comprehensive analysis of these activations from a sampling theory perspective. Our investigation reveals that, especially in shallow INRs, $\mathrm{sinc}$ activations—previously unused in conjunction with INRs—are theoretically optimal for signal encoding. Additionally, we establish a connection between dynamical systems and INRs, leveraging sampling theory to bridge these two paradigms.
|
https://proceedings.mlr.press/v235/sardana24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sardana24a/sardana24a.pdf
|
https://openreview.net/forum?id=0bmXrtTDUu
|
Beyond Chinchilla-Optimal: Accounting for Inference in Language Model Scaling Laws
|
https://proceedings.mlr.press/v235/sardana24a.html
|
Nikhil Sardana, Jacob Portes, Sasha Doubov, Jonathan Frankle
|
https://proceedings.mlr.press/v235/sardana24a.html
|
ICML 2024
|
Large language model (LLM) scaling laws are empirical formulas that estimate changes in model quality as a result of increasing parameter count and training data. However, these formulas, including the popular Deepmind Chinchilla scaling laws, neglect to include the cost of inference. We modify the Chinchilla scaling laws to calculate the optimal LLM parameter count and pre-training data size to train and deploy a model of a given quality and inference demand. We conduct our analysis both in terms of a compute budget and real-world costs and find that LLM researchers expecting reasonably large inference demand ($\sim$1B requests) should train models smaller and longer than Chinchilla-optimal. Furthermore, we train 47 models of varying sizes and parameter counts to validate our formula and find that model quality continues to improve as we scale tokens per parameter to extreme ranges (up to 10,000). Finally, we ablate the procedure used to fit the Chinchilla scaling law coefficients and find that developing scaling laws only from data collected at typical token/parameter ratios overestimates the impact of additional tokens at these extreme ranges.
|
https://proceedings.mlr.press/v235/sarfraz24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sarfraz24a/sarfraz24a.pdf
|
https://openreview.net/forum?id=W7Vqx1Jvc2
|
Position: Quo Vadis, Unsupervised Time Series Anomaly Detection?
|
https://proceedings.mlr.press/v235/sarfraz24a.html
|
M. Saquib Sarfraz, Mei-Yen Chen, Lukas Layer, Kunyu Peng, Marios Koulakis
|
https://proceedings.mlr.press/v235/sarfraz24a.html
|
ICML 2024
|
The current state of machine learning scholarship in Timeseries Anomaly Detection (TAD) is plagued by the persistent use of flawed evaluation metrics, inconsistent benchmarking practices, and a lack of proper justification for the choices made in novel deep learning-based model designs. Our paper presents a critical analysis of the status quo in TAD, revealing the misleading track of current research and highlighting problematic methods, and evaluation practices. Our position advocates for a shift in focus from solely pursuing novel model designs to improving benchmarking practices, creating non-trivial datasets, and critically evaluating the utility of complex methods against simpler baselines. Our findings demonstrate the need for rigorous evaluation protocols, the creation of simple baselines, and the revelation that state-of-the-art deep anomaly detection models effectively learn linear mappings. These findings suggest the need for more exploration and development of simple and interpretable TAD methods. The increment of model complexity in the state-of-the-art deep-learning based models unfortunately offers very little improvement. We offer insights and suggestions for the field to move forward.
|
https://proceedings.mlr.press/v235/scellier24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/scellier24a/scellier24a.pdf
|
https://openreview.net/forum?id=nAbfF37H6t
|
A fast algorithm to simulate nonlinear resistive networks
|
https://proceedings.mlr.press/v235/scellier24a.html
|
Benjamin Scellier
|
https://proceedings.mlr.press/v235/scellier24a.html
|
ICML 2024
|
Analog electrical networks have long been investigated as energy-efficient computing platforms for machine learning, leveraging analog physics during inference. More recently, resistor networks have sparked particular interest due to their ability to learn using local rules (such as equilibrium propagation), enabling potentially important energy efficiency gains for training as well. Despite their potential advantage, the simulations of these resistor networks has been a significant bottleneck to assess their scalability, with current methods either being limited to linear networks or relying on realistic, yet slow circuit simulators like SPICE. Assuming ideal circuit elements, we introduce a novel approach for the simulation of nonlinear resistive networks, which we frame as a quadratic programming problem with linear inequality constraints, and which we solve using a fast, exact coordinate descent algorithm. Our simulation methodology significantly outperforms existing SPICE-based simulations, enabling the training of networks up to 327 times larger at speeds 160 times faster, resulting in a 50,000-fold improvement in the ratio of network size to epoch duration. Our approach can foster more rapid progress in the simulations of nonlinear analog electrical networks.
|
https://proceedings.mlr.press/v235/scetbon24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/scetbon24a/scetbon24a.pdf
|
https://openreview.net/forum?id=JpzIGzru5F
|
A Fixed-Point Approach for Causal Generative Modeling
|
https://proceedings.mlr.press/v235/scetbon24a.html
|
Meyer Scetbon, Joel Jennings, Agrin Hilmkil, Cheng Zhang, Chao Ma
|
https://proceedings.mlr.press/v235/scetbon24a.html
|
ICML 2024
|
We propose a novel formalism for describing Structural Causal Models (SCMs) as fixed-point problems on causally ordered variables, eliminating the need for Directed Acyclic Graphs (DAGs), and establish the weakest known conditions for their unique recovery given the topological ordering (TO). Based on this, we design a two-stage causal generative model that first infers in a zero-shot manner a valid TO from observations, and then learns the generative SCM on the ordered variables. To infer TOs, we propose to amortize the learning of TOs on synthetically generated datasets by sequentially predicting the leaves of graphs seen during training. To learn SCMs, we design a transformer-based architecture that exploits a new attention mechanism enabling the modeling of causal structures, and show that this parameterization is consistent with our formalism. Finally, we conduct an extensive evaluation of each method individually, and show that when combined, our model outperforms various baselines on generated out-of-distribution problems.
|
https://proceedings.mlr.press/v235/schaipp24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schaipp24a/schaipp24a.pdf
|
https://openreview.net/forum?id=WvvkbWD1vL
|
MoMo: Momentum Models for Adaptive Learning Rates
|
https://proceedings.mlr.press/v235/schaipp24a.html
|
Fabian Schaipp, Ruben Ohana, Michael Eickenberg, Aaron Defazio, Robert M. Gower
|
https://proceedings.mlr.press/v235/schaipp24a.html
|
ICML 2024
|
Training a modern machine learning architecture on a new task requires extensive learning-rate tuning, which comes at a high computational cost. Here we develop new Polyak-type adaptive learning rates that can be used on top of any momentum method, and require less tuning to perform well. We first develop MoMo, a Momentum Model based adaptive learning rate for SGD-M (stochastic gradient descent with momentum). MoMo uses momentum estimates of the batch losses and gradients sampled at each iteration to build a model of the loss function. Our model also makes use of any known lower bound of the loss function by using truncation, e.g. most losses are lower-bounded by zero. The models is then approximately minimized at each iteration to compute the next step. We show how MoMo can be used in combination with any momentum-based method, and showcase this by developing MoMo-Adam - which is Adam with our new model-based adaptive learning rate. We show that MoMo attains a $\mathcal{O}(1/\sqrt{K})$ convergence rate for convex problems with interpolation, needing knowledge of no problem-specific quantities other than the optimal value. Additionally, for losses with unknown lower bounds, we develop on-the-fly estimates of a lower bound, that are incorporated in our model. We demonstrate that MoMo and MoMo-Adam improve over SGD-M and Adam in terms of robustness to hyperparameter tuning for training image classifiers on MNIST, CIFAR, and Imagenet, for recommender systems on the Criteo dataset, for a transformer model on the translation task IWSLT14, and for a diffusion model.
|
https://proceedings.mlr.press/v235/schar24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schar24a/schar24a.pdf
|
https://openreview.net/forum?id=SAXp5dMYv7
|
Parallel Affine Transformation Tuning of Markov Chain Monte Carlo
|
https://proceedings.mlr.press/v235/schar24a.html
|
Philip Schär, Michael Habeck, Daniel Rudolf
|
https://proceedings.mlr.press/v235/schar24a.html
|
ICML 2024
|
The performance of Markov chain Monte Carlo samplers strongly depends on the properties of the target distribution such as its covariance structure, the location of its probability mass and its tail behavior. We explore the use of bijective affine transformations of the sample space to improve the properties of the target distribution and thereby the performance of samplers running in the transformed space. In particular, we propose a flexible and user-friendly scheme for adaptively learning the affine transformation during sampling. Moreover, the combination of our scheme with Gibbsian polar slice sampling is shown to produce samples of high quality at comparatively low computational cost in several settings based on real-world data.
|
https://proceedings.mlr.press/v235/scheid24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/scheid24a/scheid24a.pdf
|
https://openreview.net/forum?id=ykgZk6vFrh
|
Incentivized Learning in Principal-Agent Bandit Games
|
https://proceedings.mlr.press/v235/scheid24a.html
|
Antoine Scheid, Daniil Tiapkin, Etienne Boursier, Aymeric Capitaine, Eric Moulines, Michael Jordan, El-Mahdi El-Mhamdi, Alain Oliviero Durmus
|
https://proceedings.mlr.press/v235/scheid24a.html
|
ICML 2024
|
This work considers a repeated principal-agent bandit game, where the principal can only interact with her environment through the agent. The principal and the agent have misaligned objectives and the choice of action is only left to the agent. However, the principal can influence the agent’s decisions by offering incentives which add up to his rewards. The principal aims to iteratively learn an incentive policy to maximize her own total utility. This framework extends usual bandit problems and is motivated by several practical applications, such as healthcare or ecological taxation, where traditionally used mechanism design theories often overlook the learning aspect of the problem. We present nearly optimal (with respect to a horizon $T$) learning algorithms for the principal’s regret in both multi-armed and linear contextual settings. Finally, we support our theoretical guarantees through numerical experiments.
|
https://proceedings.mlr.press/v235/schiff24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schiff24a/schiff24a.pdf
|
https://openreview.net/forum?id=mk3A5IUdn8
|
Caduceus: Bi-Directional Equivariant Long-Range DNA Sequence Modeling
|
https://proceedings.mlr.press/v235/schiff24a.html
|
Yair Schiff, Chia Hsiang Kao, Aaron Gokaslan, Tri Dao, Albert Gu, Volodymyr Kuleshov
|
https://proceedings.mlr.press/v235/schiff24a.html
|
ICML 2024
|
Large-scale sequence modeling has sparked rapid advances that now extend into biology and genomics. However, modeling genomic sequences introduces challenges such as the need to model long-range token interactions, the effects of upstream and downstream regions of the genome, and the reverse complementarity (RC) of DNA. Here, we propose an architecture motivated by these challenges that builds off the long-range Mamba block, and extends it to a BiMamba component that supports bi-directionality, and to a MambaDNA block that additionally supports RC equivariance. We use MambaDNA as the basis of Caduceus, the first family of RC equivariant bi-directional long-range DNA language models, and we introduce pre-training and fine-tuning strategies that yield Caduceus DNA foundation models. Caduceus outperforms previous long-range models on downstream benchmarks; on a challenging long-range variant effect prediction task, Caduceus exceeds the performance of 10x larger models that do not leverage bi-directionality or equivariance. Code to reproduce our experiments is available here: https://github.com/kuleshov-group/caduceus.
|
https://proceedings.mlr.press/v235/schiff24b.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schiff24b/schiff24b.pdf
|
https://openreview.net/forum?id=3abgRKnK1W
|
DySLIM: Dynamics Stable Learning by Invariant Measure for Chaotic Systems
|
https://proceedings.mlr.press/v235/schiff24b.html
|
Yair Schiff, Zhong Yi Wan, Jeffrey B. Parker, Stephan Hoyer, Volodymyr Kuleshov, Fei Sha, Leonardo Zepeda-Núñez
|
https://proceedings.mlr.press/v235/schiff24b.html
|
ICML 2024
|
Learning dynamics from dissipative chaotic systems is notoriously difficult due to their inherent instability, as formalized by their positive Lyapunov exponents, which exponentially amplify errors in the learned dynamics. However, many of these systems exhibit ergodicity and an attractor: a compact and highly complex manifold, to which trajectories converge in finite-time, that supports an invariant measure, i.e., a probability distribution that is invariant under the action of the dynamics, which dictates the long-term statistical behavior of the system. In this work, we leverage this structure to propose a new framework that targets learning the invariant measure as well as the dynamics, in contrast with typical methods that only target the misfit between trajectories, which often leads to divergence as the trajectories’ length increases. We use our framework to propose a tractable and sample efficient objective that can be used with any existing learning objectives. Our Dynamics Stable Learning by Invariant Measure (DySLIM) objective enables model training that achieves better point-wise tracking and long-term statistical accuracy relative to other learning objectives. By targeting the distribution with a scalable regularization term, we hope that this approach can be extended to more complex systems exhibiting slowly-variant distributions, such as weather and climate models. Code to reproduce our experiments is available here: https://github.com/google-research/swirl-dynamics/tree/main/swirl_dynamics/projects/ergodic.
|
https://proceedings.mlr.press/v235/schlarmann24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schlarmann24a/schlarmann24a.pdf
|
https://openreview.net/forum?id=WLPhywf1si
|
Robust CLIP: Unsupervised Adversarial Fine-Tuning of Vision Embeddings for Robust Large Vision-Language Models
|
https://proceedings.mlr.press/v235/schlarmann24a.html
|
Christian Schlarmann, Naman Deep Singh, Francesco Croce, Matthias Hein
|
https://proceedings.mlr.press/v235/schlarmann24a.html
|
ICML 2024
|
Multi-modal foundation models like OpenFlamingo, LLaVA, and GPT-4 are increasingly used for various real-world tasks. Prior work has shown that these models are highly vulnerable to adversarial attacks on the vision modality. These attacks can be leveraged to spread fake information or defraud users, and thus pose a significant risk, which makes the robustness of large multi-modal foundation models a pressing problem. The CLIP model, or one of its variants, is used as a frozen vision encoder in many large vision-language models (LVLMs), e.g. LLaVA and OpenFlamingo. We propose an unsupervised adversarial fine-tuning scheme to obtain a robust CLIP vision encoder, which yields robustness on all vision down-stream tasks (LVLMs, zero-shot classification) that rely on CLIP. In particular, we show that stealth-attacks on users of LVLMs by a malicious third party providing manipulated images are no longer possible once one replaces the original CLIP model with our robust one. No retraining or fine-tuning of the down-stream LVLMs is required. The code and robust models are available on GitHub.
|
https://proceedings.mlr.press/v235/schmidt24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schmidt24a/schmidt24a.pdf
|
https://openreview.net/forum?id=5PqzKxmfag
|
Tilt your Head: Activating the Hidden Spatial-Invariance of Classifiers
|
https://proceedings.mlr.press/v235/schmidt24a.html
|
Johann Schmidt, Sebastian Stober
|
https://proceedings.mlr.press/v235/schmidt24a.html
|
ICML 2024
|
Deep neural networks are applied in more and more areas of everyday life. However, they still lack essential abilities, such as robustly dealing with spatially transformed input signals. Approaches to mitigate this severe robustness issue are limited to two pathways: Either models are implicitly regularised by increased sample variability (data augmentation) or explicitly constrained by hard-coded inductive biases. The limiting factor of the former is the size of the data space, which renders sufficient sample coverage intractable. The latter is limited by the engineering effort required to develop such inductive biases for every possible scenario. Instead, we take inspiration from human behaviour, where percepts are modified by mental or physical actions during inference. We propose a novel technique to emulate such an inference process for neural nets. This is achieved by traversing a sparsified inverse transformation tree during inference using parallel energy-based evaluations. Our proposed inference algorithm, called Inverse Transformation Search (ITS), is model-agnostic and equips the model with zero-shot pseudo-invariance to spatially transformed inputs. We evaluated our method on several benchmark datasets, including a synthesised ImageNet test set. ITS outperforms the utilised baselines on all zero-shot test scenarios.
|
https://proceedings.mlr.press/v235/schmitt24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schmitt24a/schmitt24a.pdf
|
https://openreview.net/forum?id=6wVlH96oMX
|
Leveraging Self-Consistency for Data-Efficient Amortized Bayesian Inference
|
https://proceedings.mlr.press/v235/schmitt24a.html
|
Marvin Schmitt, Desi R. Ivanova, Daniel Habermann, Ullrich Koethe, Paul-Christian Bürkner, Stefan T. Radev
|
https://proceedings.mlr.press/v235/schmitt24a.html
|
ICML 2024
|
We propose a method to improve the efficiency and accuracy of amortized Bayesian inference by leveraging universal symmetries in the joint probabilistic model of parameters and data. In a nutshell, we invert Bayes’ theorem and estimate the marginal likelihood based on approximate representations of the joint model. Upon perfect approximation, the marginal likelihood is constant across all parameter values by definition. However, errors in approximate inference lead to undesirable variance in the marginal likelihood estimates across different parameter values. We penalize violations of this symmetry with a self-consistency loss which significantly improves the quality of approximate inference in low data regimes and can be used to augment the training of popular neural density estimators. We apply our method to a number of synthetic problems and realistic scientific models, discovering notable advantages in the context of both neural posterior and likelihood approximation.
|
https://proceedings.mlr.press/v235/schmitt-forster24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schmitt-forster24a/schmitt-forster24a.pdf
|
https://openreview.net/forum?id=07f24ya6eX
|
Regularized Q-learning through Robust Averaging
|
https://proceedings.mlr.press/v235/schmitt-forster24a.html
|
Peter Schmitt-Förster, Tobias Sutter
|
https://proceedings.mlr.press/v235/schmitt-forster24a.html
|
ICML 2024
|
We propose a new Q-learning variant, called 2RA Q-learning, that addresses some weaknesses of existing Q-learning methods in a principled manner. One such weakness is an underlying estimation bias which cannot be controlled and often results in poor performance. We propose a distributionally robust estimator for the maximum expected value term, which allows us to precisely control the level of estimation bias introduced. The distributionally robust estimator admits a closed-form solution such that the proposed algorithm has a computational cost per iteration comparable to Watkins’ Q-learning. For the tabular case, we show that 2RA Q-learning converges to the optimal policy and analyze its asymptotic mean-squared error. Lastly, we conduct numerical experiments for various settings, which corroborate our theoretical findings and indicate that 2RA Q-learning often performs better than existing methods.
|
https://proceedings.mlr.press/v235/schneider24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schneider24a/schneider24a.pdf
|
https://openreview.net/forum?id=IaV6AgrTUp
|
Implicit Representations for Constrained Image Segmentation
|
https://proceedings.mlr.press/v235/schneider24a.html
|
Jan Philipp Schneider, Mishal Fatima, Jovita Lukasik, Andreas Kolb, Margret Keuper, Michael Moeller
|
https://proceedings.mlr.press/v235/schneider24a.html
|
ICML 2024
|
Implicit representations allow to use a parametric function that maps (spatial) coordinates to the value that is traditionally stored in each pixel, e.g. RGB values, instead of a discrete grid. This has recently proven quite advantageous as an internal representation for images or scenes for deep learning models. Yet, its potential to ensure certain properties of the solution has not yet been fully explored. In this work, we demonstrate that implicit representations are a powerful tool for enforcing a variety of different geometric constraints in image segmentation. While convexity, star-shape, path-connectedness, periodicity, or symmetry of the (spatial or space-time) region to be segmented are very challenging to enforce for pixel-wise discretizations, a suitable parametrization of an implicit representation, mapping spatial or spatio-temporal coordinates to the likeliness of a pixel belonging to the fore- or background, allows to provably ensure such constraints. Several numerical examples demonstrate that challenging segmentation scenarios can benefit from the inclusion of application-specific constraints, e.g. when occlusions prevent a faithful segmentation with classical approaches.
|
https://proceedings.mlr.press/v235/schneider24b.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schneider24b/schneider24b.pdf
|
https://openreview.net/forum?id=4pFgOzKF76
|
Online Learning with Bounded Recall
|
https://proceedings.mlr.press/v235/schneider24b.html
|
Jon Schneider, Kiran Vodrahalli
|
https://proceedings.mlr.press/v235/schneider24b.html
|
ICML 2024
|
We study the problem of full-information online learning in the “bounded recall” setting popular in the study of repeated games. An online learning algorithm $\mathcal{A}$ is $M$-bounded-recall if its output at time $t$ can be written as a function of the $M$ previous rewards (and not e.g. any other internal state of $\mathcal{A}$). We first demonstrate that a natural approach to constructing bounded-recall algorithms from mean-based no-regret learning algorithms (e.g., running Hedge over the last $M$ rounds) fails, and that any such algorithm incurs constant regret per round. We then construct a stationary bounded-recall algorithm that achieves a per-round regret of $\Theta(1/\sqrt{M})$, which we complement with a tight lower bound. Finally, we show that unlike the perfect recall setting, any low regret bound bounded-recall algorithm must be aware of the ordering of the past $M$ losses – any bounded-recall algorithm which plays a symmetric function of the past $M$ losses must incur constant regret per round.
|
https://proceedings.mlr.press/v235/schopmans24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schopmans24a/schopmans24a.pdf
|
https://openreview.net/forum?id=N3ZrpSCJcJ
|
Conditional Normalizing Flows for Active Learning of Coarse-Grained Molecular Representations
|
https://proceedings.mlr.press/v235/schopmans24a.html
|
Henrik Schopmans, Pascal Friederich
|
https://proceedings.mlr.press/v235/schopmans24a.html
|
ICML 2024
|
Efficient sampling of the Boltzmann distribution of molecular systems is a long-standing challenge. Recently, instead of generating long molecular dynamics simulations, generative machine learning methods such as normalizing flows have been used to learn the Boltzmann distribution directly, without samples. However, this approach is susceptible to mode collapse and thus often does not explore the full configurational space. In this work, we address this challenge by separating the problem into two levels, the fine-grained and coarse-grained degrees of freedom. A normalizing flow conditioned on the coarse-grained space yields a probabilistic connection between the two levels. To explore the configurational space, we employ coarse-grained simulations with active learning which allows us to update the flow and make all-atom potential energy evaluations only when necessary. Using alanine dipeptide as an example, we show that our methods obtain a speedup to molecular dynamics simulations of approximately $15.9$ to $216.2$ compared to the speedup of $4.5$ of the current state-of-the-art machine learning approach.
|
https://proceedings.mlr.press/v235/schramm24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schramm24a/schramm24a.pdf
|
https://openreview.net/forum?id=UCKFhc9SFC
|
Provably Efficient Long-Horizon Exploration in Monte Carlo Tree Search through State Occupancy Regularization
|
https://proceedings.mlr.press/v235/schramm24a.html
|
Liam Schramm, Abdeslam Boularias
|
https://proceedings.mlr.press/v235/schramm24a.html
|
ICML 2024
|
Monte Carlo tree search (MCTS) has been successful in a variety of domains, but faces challenges with long-horizon exploration when compared to sampling-based motion planning algorithms like Rapidly-Exploring Random Trees. To address these limitations of MCTS, we derive a tree search algorithm based on policy optimization with state-occupancy measure regularization, which we call Volume-MCTS. We show that count-based exploration and sampling-based motion planning can be derived as approximate solutions to this state-occupancy measure regularized objective. We test our method on several robot navigation problems, and find that Volume-MCTS outperforms AlphaZero and displays significantly better long-horizon exploration properties.
|
https://proceedings.mlr.press/v235/schroder24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schroder24a/schroder24a.pdf
|
https://openreview.net/forum?id=RI4GA8amUI
|
Asymptotics of Learning with Deep Structured (Random) Features
|
https://proceedings.mlr.press/v235/schroder24a.html
|
Dominik Schröder, Daniil Dmitriev, Hugo Cui, Bruno Loureiro
|
https://proceedings.mlr.press/v235/schroder24a.html
|
ICML 2024
|
For a large class of feature maps we provide a tight asymptotic characterisation of the test error associated with learning the readout layer, in the high-dimensional limit where the input dimension, hidden layer widths, and number of training samples are proportionally large. This characterization is formulated in terms of the population covariance of the features. Our work is partially motivated by the problem of learning with Gaussian rainbow neural networks, namely deep non-linear fully-connected networks with random but structured weights, whose row-wise covariances are further allowed to depend on the weights of previous layers. For such networks we also derive a closed-form formula for the feature covariance in terms of the weight matrices. We further find that in some cases our results can capture feature maps learned by deep, finite-width neural networks trained under gradient descent.
|
https://proceedings.mlr.press/v235/schroder24b.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schroder24b/schroder24b.pdf
|
https://openreview.net/forum?id=xC7SYAZygF
|
Simultaneous identification of models and parameters of scientific simulators
|
https://proceedings.mlr.press/v235/schroder24b.html
|
Cornelius Schröder, Jakob H. Macke
|
https://proceedings.mlr.press/v235/schroder24b.html
|
ICML 2024
|
Many scientific models are composed of multiple discrete components, and scientists often make heuristic decisions about which components to include. Bayesian inference provides a mathematical framework for systematically selecting model components, but defining prior distributions over model components and developing associated inference schemes has been challenging. We approach this problem in a simulation-based inference framework: We define model priors over candidate components and, from model simulations, train neural networks to infer joint probability distributions over both model components and associated parameters. Our method, simulation-based model inference (SBMI), represents distributions over model components as a conditional mixture of multivariate binary distributions in the Grassmann formalism. SBMI can be applied to any compositional stochastic simulator without requiring likelihood evaluations. We evaluate SBMI on a simple time series model and on two scientific models from neuroscience, and show that it can discover multiple data-consistent model configurations, and that it reveals non-identifiable model components and parameters. SBMI provides a powerful tool for data-driven scientific inquiry which will allow scientists to identify essential model components and make uncertainty-informed modelling decisions.
|
https://proceedings.mlr.press/v235/schubert24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schubert24a/schubert24a.pdf
|
https://openreview.net/forum?id=BNAvYSCrLD
|
In-Context Learning Agents Are Asymmetric Belief Updaters
|
https://proceedings.mlr.press/v235/schubert24a.html
|
Johannes A. Schubert, Akshay Kumar Jagadish, Marcel Binz, Eric Schulz
|
https://proceedings.mlr.press/v235/schubert24a.html
|
ICML 2024
|
We study the in-context learning dynamics of large language models (LLMs) using three instrumental learning tasks adapted from cognitive psychology. We find that LLMs update their beliefs in an asymmetric manner and learn more from better-than-expected outcomes than from worse-than-expected ones. Furthermore, we show that this effect reverses when learning about counterfactual feedback and disappears when no agency is implied. We corroborate these findings by investigating idealized in-context learning agents derived through meta-reinforcement learning, where we observe similar patterns. Taken together, our results contribute to our understanding of how in-context learning works by highlighting that the framing of a problem significantly influences how learning occurs, a phenomenon also observed in human cognition.
|
https://proceedings.mlr.press/v235/schurholt24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schurholt24a/schurholt24a.pdf
|
https://openreview.net/forum?id=ug2uoAZ9c2
|
Towards Scalable and Versatile Weight Space Learning
|
https://proceedings.mlr.press/v235/schurholt24a.html
|
Konstantin Schürholt, Michael W. Mahoney, Damian Borth
|
https://proceedings.mlr.press/v235/schurholt24a.html
|
ICML 2024
|
Learning representations of well-trained neural network models holds the promise to provide an understanding of the inner workings of those models. However, previous work has either faced limitations when processing larger networks or was task-specific to either discriminative or generative tasks. This paper introduces the SANE approach to weight-space learning. SANE overcomes previous limitations by learning task-agnostic representations of neural networks that are scalable to larger models of varying architectures and that show capabilities beyond a single task. Our method extends the idea of hyper-representations towards sequential processing of subsets of neural network weights, thus allowing one to embed larger neural networks as a set of tokens into the learned representation space. SANE reveals global model information from layer-wise embeddings, and it can sequentially generate unseen neural network models, which was unattainable with previous hyper-representation learning methods. Extensive empirical evaluation demonstrates that SANE matches or exceeds state-of-the-art performance on several weight representation learning benchmarks, particularly in initialization for new tasks and larger ResNet architectures.
|
https://proceedings.mlr.press/v235/schweisthal24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/schweisthal24a/schweisthal24a.pdf
|
https://openreview.net/forum?id=s5PLISyNyP
|
Meta-Learners for Partially-Identified Treatment Effects Across Multiple Environments
|
https://proceedings.mlr.press/v235/schweisthal24a.html
|
Jonas Schweisthal, Dennis Frauen, Mihaela Van Der Schaar, Stefan Feuerriegel
|
https://proceedings.mlr.press/v235/schweisthal24a.html
|
ICML 2024
|
Estimating the conditional average treatment effect (CATE) from observational data is relevant for many applications such as personalized medicine. Here, we focus on the widespread setting where the observational data come from multiple environments, such as different hospitals, physicians, or countries. Furthermore, we allow for violations of standard causal assumptions, namely, overlap within the environments and unconfoundedness. To this end, we move away from point identification and focus on partial identification. Specifically, we show that current assumptions from the literature on multiple environments allow us to interpret the environment as an instrumental variable (IV). This allows us to adapt bounds from the IV literature for partial identification of CATE by leveraging treatment assignment mechanisms across environments. Then, we propose different model-agnostic learners (so-called meta-learners) to estimate the bounds that can be used in combination with arbitrary machine learning models. We further demonstrate the effectiveness of our meta-learners across various experiments using both simulated and real-world data. Finally, we discuss the applicability of our meta-learners to partial identification in instrumental variable settings, such as randomized controlled trials with non-compliance.
|
https://proceedings.mlr.press/v235/scoccola24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/scoccola24a/scoccola24a.pdf
|
https://openreview.net/forum?id=ixdfvnO0uy
|
Differentiability and Optimization of Multiparameter Persistent Homology
|
https://proceedings.mlr.press/v235/scoccola24a.html
|
Luis Scoccola, Siddharth Setlur, David Loiseaux, Mathieu Carrière, Steve Oudot
|
https://proceedings.mlr.press/v235/scoccola24a.html
|
ICML 2024
|
Real-valued functions on geometric data—such as node attributes on a graph—can be optimized using descriptors from persistent homology, allowing the user to incorporate topological terms in the loss function. When optimizing a single real-valued function (the one-parameter setting), there is a canonical choice of descriptor for persistent homology: the barcode. The operation mapping a real-valued function to its barcode is differentiable almost everywhere, and the convergence of gradient descent for losses using barcodes is relatively well understood. When optimizing a vector-valued function (the multiparameter setting), there is no unique choice of descriptor for multiparameter persistent homology, and many distinct descriptors have been proposed. This calls for the development of a general framework for differentiability and optimization that applies to a wide range of multiparameter homological descriptors. In this article, we develop such a framework and show that it encompasses well-known descriptors of different flavors, such as signed barcodes and the multiparameter persistence landscape. We complement the theory with numerical experiments supporting the idea that optimizing multiparameter homological descriptors can lead to improved performances compared to optimizing one-parameter descriptors, even when using the simplest and most efficiently computable multiparameter descriptors.
|
https://proceedings.mlr.press/v235/scott24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/scott24a/scott24a.pdf
|
https://openreview.net/forum?id=01M0N8VgfB
|
Improved Modelling of Federated Datasets using Mixtures-of-Dirichlet-Multinomials
|
https://proceedings.mlr.press/v235/scott24a.html
|
Jonathan Scott, Áine Cahill
|
https://proceedings.mlr.press/v235/scott24a.html
|
ICML 2024
|
In practice, training using federated learning can be orders of magnitude slower than standard centralized training. This severely limits the amount of experimentation and tuning that can be done, making it challenging to obtain good performance on a given task. Server-side proxy data can be used to run training simulations, for instance for hyperparameter tuning. This can greatly speed up the training pipeline by reducing the number of tuning runs to be performed overall on the true clients. However, it is challenging to ensure that these simulations accurately reflect the dynamics of the real federated training. In particular, the proxy data used for simulations often comes as a single centralized dataset without a partition into distinct clients, and partitioning this data in a naive way can lead to simulations that poorly reflect real federated training. In this paper we address the challenge of how to partition centralized data in a way that reflects the statistical heterogeneity of the true federated clients. We propose a fully federated, theoretically justified, algorithm that efficiently learns the distribution of the true clients and observe improved server-side simulations when using the inferred distribution to create simulated clients from the centralized data.
|
https://proceedings.mlr.press/v235/scotti24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/scotti24a/scotti24a.pdf
|
https://openreview.net/forum?id=65XKBGH5PO
|
MindEye2: Shared-Subject Models Enable fMRI-To-Image With 1 Hour of Data
|
https://proceedings.mlr.press/v235/scotti24a.html
|
Paul Steven Scotti, Mihir Tripathy, Cesar Torrico, Reese Kneeland, Tong Chen, Ashutosh Narang, Charan Santhirasegaran, Jonathan Xu, Thomas Naselaris, Kenneth A. Norman, Tanishq Mathew Abraham
|
https://proceedings.mlr.press/v235/scotti24a.html
|
ICML 2024
|
Reconstructions of visual perception from brain activity have improved tremendously, but the practical utility of such methods has been limited. This is because such models are trained independently per subject where each subject requires dozens of hours of expensive fMRI training data to attain high-quality results. The present work showcases high-quality reconstructions using only 1 hour of fMRI training data. We pretrain our model across 7 subjects and then fine-tune on minimal data from a new subject. Our novel functional alignment procedure linearly maps all brain data to a shared-subject latent space, followed by a shared non-linear mapping to CLIP image space. We then map from CLIP space to pixel space by fine-tuning Stable Diffusion XL to accept CLIP latents as inputs instead of text. This approach improves out-of-subject generalization with limited training data and also attains state-of-the-art image retrieval and reconstruction metrics compared to single-subject approaches. MindEye2 demonstrates how accurate reconstructions of perception are possible from a single visit to the MRI facility. All code is available on Github: https://github.com/MedARC-AI/MindEyeV2
|
https://proceedings.mlr.press/v235/seedat24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/seedat24a/seedat24a.pdf
|
https://openreview.net/forum?id=9cG1oRnqNd
|
Curated LLM: Synergy of LLMs and Data Curation for tabular augmentation in low-data regimes
|
https://proceedings.mlr.press/v235/seedat24a.html
|
Nabeel Seedat, Nicolas Huynh, Boris van Breugel, Mihaela van der Schaar
|
https://proceedings.mlr.press/v235/seedat24a.html
|
ICML 2024
|
Machine Learning (ML) in low-data settings remains an underappreciated yet crucial problem. Hence, data augmentation methods to increase the sample size of datasets needed for ML are key to unlocking the transformative potential of ML in data-deprived regions and domains. Unfortunately, the limited training set constrains traditional tabular synthetic data generators in their ability to generate a large and diverse augmented dataset needed for ML tasks. To address this challenge, we introduce $\texttt{CLLM}$, which leverages the prior knowledge of Large Language Models (LLMs) for data augmentation in the low-data regime. However, not all the data generated by LLMs will improve downstream utility, as for any generative model. Consequently, we introduce a principled curation mechanism, leveraging learning dynamics, coupled with confidence and uncertainty metrics, to obtain a high-quality dataset. Empirically, on multiple real-world datasets, we demonstrate the superior performance of $\texttt{CLLM}$ in the low-data regime compared to conventional generators. Additionally, we provide insights into the LLM generation and curation mechanism, shedding light on the features that enable them to output high-quality augmented datasets.
|
https://proceedings.mlr.press/v235/sefidgaran24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sefidgaran24a/sefidgaran24a.pdf
|
https://openreview.net/forum?id=ffS0aYP6mk
|
Lessons from Generalization Error Analysis of Federated Learning: You May Communicate Less Often!
|
https://proceedings.mlr.press/v235/sefidgaran24a.html
|
Milad Sefidgaran, Romain Chor, Abdellatif Zaidi, Yijun Wan
|
https://proceedings.mlr.press/v235/sefidgaran24a.html
|
ICML 2024
|
We investigate the generalization error of statistical learning models in a Federated Learning (FL) setting. Specifically, we study the evolution of the generalization error with the number of communication rounds $R$ between $K$ clients and a parameter server (PS), i.e. the effect on the generalization error of how often the clients’ local models are aggregated at PS. In our setup, the more the clients communicate with PS the less data they use for local training in each round, such that the amount of training data per client is identical for distinct values of $R$. We establish PAC-Bayes and rate-distortion theoretic bounds on the generalization error that account explicitly for the effect of the number of rounds $R$, in addition to the number of participating devices $K$ and individual datasets size $n$. The bounds, which apply to a large class of loss functions and learning algorithms, appear to be the first of their kind for the FL setting. Furthermore, we apply our bounds to FL-type Support Vector Machines (FSVM); and derive (more) explicit bounds in this case. In particular, we show that the generalization bound of FSVM increases with $R$, suggesting that more frequent communication with PS diminishes the generalization power. This implies that the population risk decreases less fast with $R$ than does the empirical risk. Moreover, our bound suggests that the generalization error of FSVM decreases faster than that of centralized learning by a factor of $\mathcal{O}(\sqrt{\log(K)/K})$. Finally, we provide experimental results obtained using neural networks (ResNet-56) which show evidence that not only may our observations for FSVM hold more generally but also that the population risk may even start to increase beyond some value of $R$.
|
https://proceedings.mlr.press/v235/sel24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/sel24a/sel24a.pdf
|
https://openreview.net/forum?id=KJL2b6BthC
|
Algorithm of Thoughts: Enhancing Exploration of Ideas in Large Language Models
|
https://proceedings.mlr.press/v235/sel24a.html
|
Bilgehan Sel, Ahmad Tawaha, Vanshaj Khattar, Ruoxi Jia, Ming Jin
|
https://proceedings.mlr.press/v235/sel24a.html
|
ICML 2024
|
Current literature, aiming to surpass the "Chain-of-Thought" approach, often resorts to external modi operandi involving halting, modifying, and then resuming the generation process to boost Large Language Models’ (LLMs) reasoning capacities. Due to their myopic perspective, they escalate the number of query requests, leading to increased costs, memory, and computational overheads. Addressing this, we propose the Algorithm of Thoughts—a novel strategy that propels LLMs through algorithmic reasoning pathways. By employing algorithmic examples fully in-context, this overarching view of the whole process exploits the innate recurrence dynamics of LLMs, expanding their idea exploration with merely one or a few queries. Our technique outperforms earlier single-query methods and even more recent multi-query strategies that employ an extensive tree search algorithms while using significantly fewer tokens. Intriguingly, our results suggest that instructing an LLM using an algorithm can lead to performance surpassing that of the algorithm itself, hinting at LLM’s inherent ability to weave its intuition into optimized searches. We probe into the underpinnings of our method’s efficacy and its nuances in application. The code and related content can be found in: https://algorithm-of-thoughts.github.io
|
https://proceedings.mlr.press/v235/seo24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/seo24a/seo24a.pdf
|
https://openreview.net/forum?id=1xKgDANODx
|
Retrieval-Augmented Score Distillation for Text-to-3D Generation
|
https://proceedings.mlr.press/v235/seo24a.html
|
Junyoung Seo, Susung Hong, Wooseok Jang, Inès Hyeonsu Kim, Min-Seop Kwak, Doyup Lee, Seungryong Kim
|
https://proceedings.mlr.press/v235/seo24a.html
|
ICML 2024
|
Text-to-3D generation has achieved significant success by incorporating powerful 2D diffusion models, but insufficient 3D prior knowledge also leads to the inconsistency of 3D geometry. Recently, since large-scale multi-view datasets have been released, fine-tuning the diffusion model on the multi-view datasets becomes a mainstream to solve the 3D inconsistency problem. However, it has confronted with fundamental difficulties regarding the limited quality and diversity of 3D data, compared with 2D data. To sidestep these trade-offs, we explore a retrieval-augmented approach tailored for score distillation, dubbed ReDream. We postulate that both expressiveness of 2D diffusion models and geometric consistency of 3D assets can be fully leveraged by employing the semantically relevant assets directly within the optimization process. To this end, we introduce novel framework for retrieval-based quality enhancement in text-to-3D generation. We leverage the retrieved asset to incorporate its geometric prior in the variational objective and adapt the diffusion model’s 2D prior toward view consistency, achieving drastic improvements in both geometry and fidelity of generated scenes. We conduct extensive experiments to demonstrate that ReDream exhibits superior quality with increased geometric consistency. Project page is available at https://ku-cvlab.github.io/ReDream/.
|
https://proceedings.mlr.press/v235/seong24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/seong24a/seong24a.pdf
|
https://openreview.net/forum?id=dFEeI51O5j
|
Self-Supervised Interpretable End-to-End Learning via Latent Functional Modularity
|
https://proceedings.mlr.press/v235/seong24a.html
|
Hyunki Seong, Hyunchul Shim
|
https://proceedings.mlr.press/v235/seong24a.html
|
ICML 2024
|
We introduce MoNet, a novel functionally modular network for self-supervised and interpretable end-to-end learning. By leveraging its functional modularity with a latent-guided contrastive loss function, MoNet efficiently learns task-specific decision-making processes in latent space without requiring task-level supervision. Moreover, our method incorporates an online, post-hoc explainability approach that enhances the interpretability of end-to-end inferences without compromising sensorimotor control performance. In real-world indoor environments, MoNet demonstrates effective visual autonomous navigation, outperforming baseline models by 7% to 28% in task specificity analysis. We further explore the interpretability of our network through post-hoc analysis of perceptual saliency maps and latent decision vectors. This provides valuable insights into the incorporation of explainable artificial intelligence into robotic learning, encompassing both perceptual and behavioral perspectives. Supplementary materials are available at https://sites.google.com/view/monet-lgc.
|
https://proceedings.mlr.press/v235/setlur24a.html
|
https://raw.githubusercontent.com/mlresearch/v235/main/assets/setlur24a/setlur24a.pdf
|
https://openreview.net/forum?id=fdroxYsgzQ
|
Prompting is a Double-Edged Sword: Improving Worst-Group Robustness of Foundation Models
|
https://proceedings.mlr.press/v235/setlur24a.html
|
Amrith Setlur, Saurabh Garg, Virginia Smith, Sergey Levine
|
https://proceedings.mlr.press/v235/setlur24a.html
|
ICML 2024
|
Machine learning models fail catastrophically under distribution shift, but a surprisingly effective way to empirically improve robustness to some types of shift (e.g., Imagenet-A/C) is to use stronger open-vocabulary classifiers derived from foundation models. In this work, we first note that for shifts governed by spurious correlations (features spuriously correlated with the label on the training data, but not on test), the zero-shot and few-shot performance of foundation models is no better than ERM models, and remains unchanged when pretrained data/model size is scaled. Secondly, even in these situations, foundation models are quite accurate at predicting the value of the spurious feature. In a simplified setup, we theoretically analyze both these findings. Specifically, we show that during contrastive pretraining, the simplicity bias of foundation models tends to result in the learning of features that mostly rely on the spurious attribute, compared to more robust features. We leverage these observations to propose Prompting for Robustness (PfR) which first uses foundation models to zero-shot predict the spurious attribute on labeled examples, and then learns a classifier with balanced performance across different groups of labels and spurious attribute. Across 5 vision and language tasks, we show that PfR’s performance nearly equals that of an oracle algorithm (group DRO) that leverages human labeled spurious attributes.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.