title
stringlengths 7
239
| abstract
stringlengths 7
2.76k
| cs
int64 0
1
| phy
int64 0
1
| math
int64 0
1
| stat
int64 0
1
| quantitative biology
int64 0
1
| quantitative finance
int64 0
1
|
|---|---|---|---|---|---|---|---|
Metric-Optimized Example Weights | Real-world machine learning applications often have complex test metrics, and
may have training and test data that follow different distributions. We propose
addressing these issues by using a weighted loss function with a standard
convex loss, but with weights on the training examples that are learned to
optimize the test metric of interest on the validation set. These
metric-optimized example weights can be learned for any test metric, including
black box losses and customized metrics for specific applications. We
illustrate the performance of our proposal with public benchmark datasets and
real-world applications with domain shift and custom loss functions that
balance multiple objectives, impose fairness policies, and are non-convex and
non-decomposable.
| 0 | 0 | 0 | 1 | 0 | 0 |
Interior Eigensolver for Sparse Hermitian Definite Matrices Based on Zolotarev's Functions | This paper proposes an efficient method for computing selected generalized
eigenpairs of a sparse Hermitian definite matrix pencil (A, B). Based on
Zolotarev's best rational function approximations of the signum function and
conformal mapping techniques, we construct the best rational function
approximation of a rectangular function supported on an arbitrary interval.
This new best rational function approximation is applied to construct spectrum
filters of (A, B). Combining fast direct solvers and the shift-invariant GMRES,
a hybrid fast algorithm is proposed to apply spectral filters efficiently.
Compared to the state-of-the-art algorithm FEAST, the proposed rational
function approximation is proved to be optimal among a larger function class,
and the numerical implementation of the proposed method is also faster. The
efficiency and stability of the proposed method are demonstrated by numerical
examples from computational chemistry.
| 1 | 0 | 1 | 0 | 0 | 0 |
A Polya-Vinogradov-type inequality on $\mathbb{Z}[i]$ | We establish a Polya-Vinogradov-type bound for finite periodic multipicative
characters on the Gaussian integers.
| 0 | 0 | 1 | 0 | 0 | 0 |
Distinction of representations via Bruhat-Tits buildings of p-adic groups | Introductory and pedagogical treatmeant of the article : P. Broussous
"Distinction of the Steinberg representation", with an appendix by François
Courtès, IMRN 2014, no 11, 3140-3157. To appear in Proceedings of Chaire Jean
Morlet, Dipendra Prasad, Volker Heiermann Ed. 2017. Contains modified and
simplified proofs of loc. cit. This article is written in memory of
François Courtès who passed away in september 2016.
| 0 | 0 | 1 | 0 | 0 | 0 |
Face Super-Resolution Through Wasserstein GANs | Generative adversarial networks (GANs) have received a tremendous amount of
attention in the past few years, and have inspired applications addressing a
wide range of problems. Despite its great potential, GANs are difficult to
train. Recently, a series of papers (Arjovsky & Bottou, 2017a; Arjovsky et al.
2017b; and Gulrajani et al. 2017) proposed using Wasserstein distance as the
training objective and promised easy, stable GAN training across architectures
with minimal hyperparameter tuning. In this paper, we compare the performance
of Wasserstein distance with other training objectives on a variety of GAN
architectures in the context of single image super-resolution. Our results
agree that Wasserstein GAN with gradient penalty (WGAN-GP) provides stable and
converging GAN training and that Wasserstein distance is an effective metric to
gauge training progress.
| 1 | 0 | 0 | 1 | 0 | 0 |
Network-based protein structural classification | Experimental determination of protein function is resource-consuming. As an
alternative, computational prediction of protein function has received
attention. In this context, protein structural classification (PSC) can help,
by allowing for determining structural classes of currently unclassified
proteins based on their features, and then relying on the fact that proteins
with similar structures have similar functions. Existing PSC approaches rely on
sequence-based or direct ("raw") 3-dimensional (3D) structure-based protein
features. In contrast, we first model 3D structures as protein structure
networks (PSNs). Then, we use ("processed") network-based features for PSC. We
propose the use of graphlets, state-of-the-art features in many domains of
network science, in the task of PSC. Moreover, because graphlets can deal only
with unweighted PSNs, and because accounting for edge weights when constructing
PSNs could improve PSC accuracy, we also propose a deep learning framework that
automatically learns network features from the weighted PSNs. When evaluated on
a large set of ~9,400 CATH and ~12,800 SCOP protein domains (spanning 36 PSN
sets), our proposed approaches are superior to existing PSC approaches in terms
of accuracy, with comparable running time.
| 0 | 0 | 0 | 1 | 1 | 0 |
Deep Learning for Environmentally Robust Speech Recognition: An Overview of Recent Developments | Eliminating the negative effect of non-stationary environmental noise is a
long-standing research topic for automatic speech recognition that stills
remains an important challenge. Data-driven supervised approaches, including
ones based on deep neural networks, have recently emerged as potential
alternatives to traditional unsupervised approaches and with sufficient
training, can alleviate the shortcomings of the unsupervised methods in various
real-life acoustic environments. In this light, we review recently developed,
representative deep learning approaches for tackling non-stationary additive
and convolutional degradation of speech with the aim of providing guidelines
for those involved in the development of environmentally robust speech
recognition systems. We separately discuss single- and multi-channel techniques
developed for the front-end and back-end of speech recognition systems, as well
as joint front-end and back-end training frameworks.
| 1 | 0 | 0 | 0 | 0 | 0 |
The function field Sathé-Selberg formula in arithmetic progressions and `short intervals' | We use a function field analogue of a method of Selberg to derive an
asymptotic formula for the number of (square-free) monic polynomials in
$\mathbb{F}_q[X]$ of degree $n$ with precisely $k$ irreducible factors, in the
limit as $n$ tends to infinity. We then adapt this method to count such
polynomials in arithmetic progressions and short intervals, and by making use
of Weil's `Riemann hypothesis' for curves over $\mathbb{F}_q$, obtain better
ranges for these formulae than are currently known for their analogues in the
number field setting. Finally, we briefly discuss the regime in which $q$ tends
to infinity.
| 0 | 0 | 1 | 0 | 0 | 0 |
Polynomial Relations Between Matrices of Graphs | We derive a correspondence between the eigenvalues of the adjacency matrix
$A$ and the signless Laplacian matrix $Q$ of a graph $G$ when $G$ is
$(d_1,d_2)$-biregular by using the relation $A^2=(Q-d_1I)(Q-d_2I)$. This
motivates asking when it is possible to have $X^r=f(Y)$ for $f$ a polynomial,
$r>0$, and $X,\ Y$ matrices associated to a graph $G$. It turns out that,
essentially, this can only happen if $G$ is either regular or biregular.
| 0 | 0 | 1 | 0 | 0 | 0 |
Random active path model of deep neural networks with diluted binary synapses | Deep learning has become a powerful and popular tool for a variety of machine
learning tasks. However, it is challenging to understand the mechanism of deep
learning from a theoretical perspective. In this work, we propose a random
active path model to study collective properties of deep neural networks with
binary synapses, under the removal perturbation of connections between layers.
In the model, the path from input to output is randomly activated, and the
corresponding input unit constrains the weights along the path into the form of
a $p$-weight interaction glass model. A critical value of the perturbation is
observed to separate a spin glass regime from a paramagnetic regime, with the
transition being of the first order. The paramagnetic phase is conjectured to
have a poor generalization performance.
| 1 | 0 | 0 | 1 | 0 | 0 |
Super Generalized Central Limit Theorem: Limit distributions for sums of non-identical random variables with power-laws | In nature or societies, the power-law is present ubiquitously, and then it is
important to investigate the mathematical characteristics of power-laws in the
recent era of big data. In this paper we prove the superposition of
non-identical stochastic processes with power-laws converges in density to a
unique stable distribution. This property can be used to explain the
universality of stable laws such that the sums of the logarithmic return of
non-identical stock price fluctuations follow stable distributions.
| 0 | 0 | 1 | 1 | 0 | 0 |
On the Adjacency Spectra of Hypertrees | We extend the results of Zhang et al. to show that $\lambda$ is an eigenvalue
of a $k$-uniform hypertree $(k \geq 3)$ if and only if it is a root of a
particular matching polynomial for a connected induced subtree. We then use
this to provide a spectral characterization for power hypertrees. Notably, the
situation is quite different from that of ordinary trees, i.e., $2$-uniform
trees. We conclude by presenting an example (an $11$ vertex, $3$-uniform
non-power hypertree) illustrating these phenomena.
| 0 | 0 | 1 | 0 | 0 | 0 |
Coherent structures and spectral energy transfer in turbulent plasma: a space-filter approach | Plasma turbulence at scales of the order of the ion inertial length is
mediated by several mechanisms, including linear wave damping, magnetic
reconnection, formation and dissipation of thin current sheets, stochastic
heating. It is now understood that the presence of localized coherent
structures enhances the dissipation channels and the kinetic features of the
plasma. However, no formal way of quantifying the relationship between
scale-to-scale energy transfer and the presence of spatial structures has so
far been presented. In this letter we quantify such relationship analyzing the
results of a two-dimensional high-resolution Hall-MHD simulation. In
particular, we employ the technique of space-filtering to derive a spectral
energy flux term which defines, in any point of the computational domain, the
signed flux of spectral energy across a given wavenumber. The characterization
of coherent structures is performed by means of a traditional two-dimensional
wavelet transformation. By studying the correlation between the spectral energy
flux and the wavelet amplitude, we demonstrate the strong relationship between
scale-to-scale transfer and coherent structures. Furthermore, by conditioning
one quantity with respect to the other, we are able for the first time to
quantify the inhomogeneity of the turbulence cascade induced by topological
structures in the magnetic field. Taking into account the low filling-factor of
coherent structures (i.e. they cover a small portion of space), it emerges that
80% of the spectral energy transfer (both in the direct and inverse cascade
directions) is localized in about 50% of space, and 50% of the energy transfer
is localized in only 25% of space.
| 0 | 1 | 0 | 0 | 0 | 0 |
Bayesian Sparsification of Recurrent Neural Networks | Recurrent neural networks show state-of-the-art results in many text analysis
tasks but often require a lot of memory to store their weights. Recently
proposed Sparse Variational Dropout eliminates the majority of the weights in a
feed-forward neural network without significant loss of quality. We apply this
technique to sparsify recurrent neural networks. To account for recurrent
specifics we also rely on Binary Variational Dropout for RNN. We report 99.5%
sparsity level on sentiment analysis task without a quality drop and up to 87%
sparsity level on language modeling task with slight loss of accuracy.
| 1 | 0 | 0 | 1 | 0 | 0 |
Malaria Likelihood Prediction By Effectively Surveying Households Using Deep Reinforcement Learning | We build a deep reinforcement learning (RL) agent that can predict the
likelihood of an individual testing positive for malaria by asking questions
about their household. The RL agent learns to determine which survey question
to ask next and when to stop to make a prediction about their likelihood of
malaria based on their responses hitherto. The agent incurs a small penalty for
each question asked, and a large reward/penalty for making the correct/wrong
prediction; it thus has to learn to balance the length of the survey with the
accuracy of its final predictions. Our RL agent is a Deep Q-network that learns
a policy directly from the responses to the questions, with an action defined
for each possible survey question and for each possible prediction class. We
focus on Kenya, where malaria is a massive health burden, and train the RL
agent on a dataset of 6481 households from the Kenya Malaria Indicator Survey
2015. To investigate the importance of having survey questions be adaptive to
responses, we compare our RL agent to a supervised learning (SL) baseline that
fixes its set of survey questions a priori. We evaluate on prediction accuracy
and on the number of survey questions asked on a holdout set and find that the
RL agent is able to predict with 80% accuracy, using only 2.5 questions on
average. In addition, the RL agent learns to survey adaptively to responses and
is able to match the SL baseline in prediction accuracy while significantly
reducing survey length.
| 1 | 0 | 0 | 1 | 0 | 0 |
Curious Minds Wonder Alike: Studying Multimodal Behavioral Dynamics to Design Social Scaffolding of Curiosity | Curiosity is the strong desire to learn or know more about something or
someone. Since learning is often a social endeavor, social dynamics in
collaborative learning may inevitably influence curiosity. There is a scarcity
of research, however, focusing on how curiosity can be evoked in group learning
contexts. Inspired by a recently proposed theoretical framework that
articulates an integrated socio-cognitive infrastructure of curiosity, in this
work, we use data-driven approaches to identify fine-grained social scaffolding
of curiosity in child-child interaction, and propose how they can be used to
elicit and maintain curiosity in technology-enhanced learning environments. For
example, we discovered sequential patterns of multimodal behaviors across group
members and we describe those that maximize an individual's utility, or
likelihood, of demonstrating curiosity during open-ended problem-solving in
group work. We also discovered, and describe here, behaviors that directly or
in a mediated manner cause curiosity related conversational behaviors in the
interaction, with twice as many interpersonal causal influences compared to
intrapersonal ones. We explain how these findings form a solid foundation for
developing curiosity-increasing learning technologies or even assisting a human
coach to induce curiosity among learners.
| 1 | 0 | 0 | 0 | 0 | 0 |
Communication Complexity of Discrete Fair Division | We initiate the study of the communication complexity of fair division with
indivisible goods. We focus on some of the most well-studied fairness notions
(envy-freeness, proportionality, and approximations thereof) and valuation
classes (submodular, subadditive and unrestricted). Within these parameters,
our results completely resolve whether the communication complexity of
computing a fair allocation (or determining that none exist) is polynomial or
exponential (in the number of goods), for every combination of fairness notion,
valuation class, and number of players, for both deterministic and randomized
protocols.
| 1 | 0 | 0 | 0 | 0 | 0 |
A proof of the Muir-Suffridge conjecture for convex maps of the unit ball in $\mathbb C^n$ | We prove (and improve) the Muir-Suffridge conjecture for holomorphic convex
maps. Namely, let $F:\mathbb B^n\to \mathbb C^n$ be a univalent map from the
unit ball whose image $D$ is convex. Let $\mathcal S\subset \partial \mathbb
B^n$ be the set of points $\xi$ such that $\lim_{z\to \xi}\|F(z)\|=\infty$.
Then we prove that $\mathcal S$ is either empty, or contains one or two points
and $F$ extends as a homeomorphism $\tilde{F}:\overline{\mathbb B^n}\setminus
\mathcal S\to \overline{D}$. Moreover, $\mathcal S=\emptyset$ if $D$ is
bounded, $\mathcal S$ has one point if $D$ has one connected component at
$\infty$ and $\mathcal S$ has two points if $D$ has two connected components at
$\infty$ and, up to composition with an affine map, $F$ is an extension of the
strip map in the plane to higher dimension.
| 0 | 0 | 1 | 0 | 0 | 0 |
Continuous Functional Calculus for Quaternionic Bounded Normal Operators | In this article we give an approach to define continuous functional calculus
for bounded quaternionic normal operators defined on a right quaternionic
Hilbert space.
| 0 | 0 | 1 | 0 | 0 | 0 |
Evolution of structure, magnetism and electronic transport in doped pyrochlore iridate Y$_2$Ir$_{2-x}$Ru$_{x}$O$_7$ | The interplay between spin-orbit coupling (SOC) and electron correlation
($U$) is considered for many exotic phenomena in iridium oxides. We have
investigated the evolution of structural, magnetic and electronic properties in
pyrochlore iridate Y$_2$Ir$_{2-x}$Ru$_{x}$O$_7$ where the substitution of Ru
has been aimed to tune this interplay. The Ru substitution does not introduce
any structural phase transition, however, we do observe an evolution of lattice
parameters with the doping level $x$. X-ray photoemission spectroscopy (XPS)
study indicates Ru adopts charge state of Ru$^{4+}$ and replaces the Ir$^{4+}$
accordingly. Magnetization data reveal both the onset of magnetic
irreversibility and the magnetic moment decreases with progressive substitution
of Ru. These materials show non-equilibrium low temperature magnetic state as
revealed by magnetic relaxation data. Interestingly, we find magnetic
relaxation rate increases with substitution of Ru. The electrical resistivity
shows an insulating behavior in whole temperature range, however, resistivity
decreases with substitution of Ru. Nature of electronic conduction has been
found to follow power-law behavior for all the materials.
| 0 | 1 | 0 | 0 | 0 | 0 |
Cieliebak's Invariance Theorem and contact structures via connected sums | We present a strong version of Abouzaid's No-Escape Lemma, which allows
varying contact forms on the boundary and which can be used instead of the
Maximum Principle. Moreover, we give a clarified proof of Cieliebak's
Invariance Theorem for Symplectic homology under subcritical handle attachment.
Finally, we introduce the notion of asymptotically finitely generated contact
structures, which states essentially that the Symplectic homology in a certain
degree of any filling of such contact manifolds is uniformly generated by only
finitely many Reeb orbits. This property is then used to show that a large
class of manifolds carries infinitely many exactly fillable contact structures.
| 0 | 0 | 1 | 0 | 0 | 0 |
Learning Graph Weighted Models on Pictures | Graph Weighted Models (GWMs) have recently been proposed as a natural
generalization of weighted automata over strings and trees to arbitrary
families of labeled graphs (and hypergraphs). A GWM generically associates a
labeled graph with a tensor network and computes a value by successive
contractions directed by its edges. In this paper, we consider the problem of
learning GWMs defined over the graph family of pictures (or 2-dimensional
words). As a proof of concept, we consider regression and classification tasks
over the simple Bars & Stripes and Shifting Bits picture languages and provide
an experimental study investigating whether these languages can be learned in
the form of a GWM from positive and negative examples using gradient-based
methods. Our results suggest that this is indeed possible and that
investigating the use of gradient-based methods to learn picture series and
functions computed by GWMs over other families of graphs could be a fruitful
direction.
| 0 | 0 | 0 | 1 | 0 | 0 |
Solutions for biharmonic equations with steep potential wells | In this paper, we are concerned with the existence of least energy solutions
for the following biharmonic equations: $$\Delta^2 u+(\lambda
V(x)-\delta)u=|u|^{p-2}u \quad in\quad \mathbb{R}^N$$ where $N\geq 5,
2<p\leq\frac{2N}{N-4}, \lambda>0$ is a parameter, $V(x)$ is a nonnegative
potential function with nonempty zero sets $\mbox{int} V^{-1}(0)$,
$0<\delta<\mu_0$ and $\mu_0$ is the principle eigenvalue of $\Delta^2$ in the
zero sets $\mbox{int} V^{-1}(0)$ of $V(x)$. Here $\mbox{int} V^{-1}(0)$ denotes
the interior part of the set $V^{-1}(0):=\{x\in \mathbb{R}^N: V(x)=0\}$. We
prove that the above equation admits a least energy solution which is trapped
near the zero sets $\mbox{int} V^{-1}(0)$ for $\lambda>0$ large.
| 0 | 0 | 1 | 0 | 0 | 0 |
General three and four person two color Hat Game | N distinguishable players are randomly fitted with a white or black hat,
where the probabilities of getting a white or black hat may be different for
each player, but known to all the players. All players guess simultaneously the
color of their own hat observing only the hat colors of the other N-1 players.
It is also allowed for each player to pass: no color is guessed. The team wins
if at least one player guesses his hat color correctly and none of the players
has an incorrect guess. No communication of any sort is allowed, except for an
initial strategy session before the game begins. Our goal is to maximize the
probability of winning the game and to describe winning strategies, using the
concept of an adequate set. We find explicit solutions in case of N =3 and N
=4.
| 1 | 0 | 0 | 0 | 0 | 0 |
Pencilled regular parallelisms | Over any field $\mathbb K$, there is a bijection between regular spreads of
the projective space ${\rm PG}(3,{\mathbb K})$ and $0$-secant lines of the
Klein quadric in ${\rm PG}(5,{\mathbb K})$. Under this bijection, regular
parallelisms of ${\rm PG}(3,{\mathbb K})$ correspond to hyperflock determining
line sets (hfd line sets) with respect to the Klein quadric. An hfd line set is
defined to be \emph{pencilled} if it is composed of pencils of lines. We
present a construction of pencilled hfd line sets, which is then shown to
determine all such sets. Based on these results, we describe the corresponding
regular parallelisms. These are also termed as being \emph{pencilled}. Any
Clifford parallelism is regular and pencilled. From this, we derive necessary
and sufficient algebraic conditions for the existence of pencilled hfd line
sets.
| 0 | 0 | 1 | 0 | 0 | 0 |
The loss surface of deep and wide neural networks | While the optimization problem behind deep neural networks is highly
non-convex, it is frequently observed in practice that training deep networks
seems possible without getting stuck in suboptimal points. It has been argued
that this is the case as all local minima are close to being globally optimal.
We show that this is (almost) true, in fact almost all local minima are
globally optimal, for a fully connected network with squared loss and analytic
activation function given that the number of hidden units of one layer of the
network is larger than the number of training points and the network structure
from this layer on is pyramidal.
| 1 | 0 | 0 | 1 | 0 | 0 |
Gap Acceptance During Lane Changes by Large-Truck Drivers-An Image-Based Analysis | This paper presents an analysis of rearward gap acceptance characteristics of
drivers of large trucks in highway lane change scenarios. The range between the
vehicles was inferred from camera images using the estimated lane width
obtained from the lane tracking camera as the reference. Six-hundred lane
change events were acquired from a large-scale naturalistic driving data set.
The kinematic variables from the image-based gap analysis were filtered by the
weighted linear least squares in order to extrapolate them at the lane change
time. In addition, the time-to-collision and required deceleration were
computed, and potential safety threshold values are provided. The resulting
range and range rate distributions showed directional discrepancies, i.e., in
left lane changes, large trucks are often slower than other vehicles in the
target lane, whereas they are usually faster in right lane changes. Video
observations have confirmed that major motivations for changing lanes are
different depending on the direction of move, i.e., moving to the left (faster)
lane occurs due to a slower vehicle ahead or a merging vehicle on the
right-hand side, whereas right lane changes are frequently made to return to
the original lane after passing.
| 1 | 0 | 0 | 0 | 0 | 0 |
Diversity-Sensitive Conditional Generative Adversarial Networks | We propose a simple yet highly effective method that addresses the
mode-collapse problem in the Conditional Generative Adversarial Network (cGAN).
Although conditional distributions are multi-modal (i.e., having many modes) in
practice, most cGAN approaches tend to learn an overly simplified distribution
where an input is always mapped to a single output regardless of variations in
latent code. To address such issue, we propose to explicitly regularize the
generator to produce diverse outputs depending on latent codes. The proposed
regularization is simple, general, and can be easily integrated into most
conditional GAN objectives. Additionally, explicit regularization on generator
allows our method to control a balance between visual quality and diversity. We
demonstrate the effectiveness of our method on three conditional generation
tasks: image-to-image translation, image inpainting, and future video
prediction. We show that simple addition of our regularization to existing
models leads to surprisingly diverse generations, substantially outperforming
the previous approaches for multi-modal conditional generation specifically
designed in each individual task.
| 1 | 0 | 0 | 1 | 0 | 0 |
Predicting Expressive Speaking Style From Text In End-To-End Speech Synthesis | Global Style Tokens (GSTs) are a recently-proposed method to learn latent
disentangled representations of high-dimensional data. GSTs can be used within
Tacotron, a state-of-the-art end-to-end text-to-speech synthesis system, to
uncover expressive factors of variation in speaking style. In this work, we
introduce the Text-Predicted Global Style Token (TP-GST) architecture, which
treats GST combination weights or style embeddings as "virtual" speaking style
labels within Tacotron. TP-GST learns to predict stylistic renderings from text
alone, requiring neither explicit labels during training nor auxiliary inputs
for inference. We show that, when trained on a dataset of expressive speech,
our system generates audio with more pitch and energy variation than two
state-of-the-art baseline models. We further demonstrate that TP-GSTs can
synthesize speech with background noise removed, and corroborate these analyses
with positive results on human-rated listener preference audiobook tasks.
Finally, we demonstrate that multi-speaker TP-GST models successfully factorize
speaker identity and speaking style. We provide a website with audio samples
for each of our findings.
| 1 | 0 | 0 | 1 | 0 | 0 |
Independence of Sources in Social Networks | Online social networks are more and more studied. The links between users of
a social network are important and have to be well qualified in order to detect
communities and find influencers for example. In this paper, we present an
approach based on the theory of belief functions to estimate the degrees of
cognitive independence between users in a social network. We experiment the
proposed method on a large amount of data gathered from the Twitter social
network.
| 1 | 0 | 0 | 0 | 0 | 0 |
Classical and quantum systems: transport due to rare events | We review possible mechanisms for energy transfer based on 'rare' or
'non-perturbative' effects, in physical systems that present a many-body
localized phenomenology. The main focus is on classical systems, with or
without quenched disorder. For non-quantum systems, the breakdown of
localization is usually not regarded as an issue, and we thus aim at
identifying the fastest channels for transport. Next, we contemplate the
possibility of applying the same mechanisms in quantum systems, including
disorder free systems (e.g. Bose-Hubbard chain), disordered many-body localized
systems with mobility edges at energies below the edge, and strongly disordered
lattice systems in $d>1$. For quantum mechanical systems, the relevance of
these considerations for transport is currently a matter of debate.
| 0 | 1 | 0 | 0 | 0 | 0 |
An a Priori Exponential Tail Bound for k-Folds Cross-Validation | We consider a priori generalization bounds developed in terms of
cross-validation estimates and the stability of learners. In particular, we
first derive an exponential Efron-Stein type tail inequality for the
concentration of a general function of n independent random variables. Next,
under some reasonable notion of stability, we use this exponential tail bound
to analyze the concentration of the k-fold cross-validation (KFCV) estimate
around the true risk of a hypothesis generated by a general learning rule.
While the accumulated literature has often attributed this concentration to the
bias and variance of the estimator, our bound attributes this concentration to
the stability of the learning rule and the number of folds k. This insight
raises valid concerns related to the practical use of KFCV and suggests
research directions to obtain reliable empirical estimates of the actual risk.
| 1 | 0 | 0 | 1 | 0 | 0 |
Relaxed Wasserstein with Applications to GANs | We propose a novel class of statistical divergences called \textit{Relaxed
Wasserstein} (RW) divergence. RW divergence generalizes Wasserstein divergence
and is parametrized by a class of strictly convex and differentiable functions.
We establish for RW divergence several probabilistic properties, which are
critical for the success of Wasserstein divergence. In particular, we show that
RW divergence is dominated by Total Variation (TV) and Wasserstein-$L^2$
divergence, and that RW divergence has continuity, differentiability and
duality representation. Finally, we provide a nonasymptotic moment estimate and
a concentration inequality for RW divergence.
Our experiments on the image generation task demonstrate that RW divergence
is a suitable choice for GANs. Indeed, the performance of RWGANs with
Kullback-Leibler (KL) divergence is very competitive with other
state-of-the-art GANs approaches. Furthermore, RWGANs possess better
convergence properties than the existing WGANs with competitive inception
scores. To the best of our knowledge, our new conceptual framework is the first
to not only provide the flexibility in designing effective GANs scheme, but
also the possibility in studying different losses under a unified mathematical
framework.
| 1 | 0 | 0 | 1 | 0 | 0 |
Extremal functions for the Moser--Trudinger inequality of Adimurthi--Druet type in $W^{1,N}(\mathbb R^N)$ | We study the existence and nonexistence of maximizers for variational problem
concerning to the Moser--Trudinger inequality of Adimurthi--Druet type in
$W^{1,N}(\mathbb R^N)$ \[ MT(N,\beta, \alpha) =\sup_{u\in W^{1,N}(\mathbb R^N),
\|\nabla u\|_N^N + \|u\|_N^N\leq 1} \int_{\mathbb R^N} \Phi_N(\beta(1+\alpha
\|u\|_N^N)^{\frac1{N-1}} |u|^{\frac N{N-1}}) dx, \] where $\Phi_N(t) =e^{t}
-\sum_{k=0}^{N-2} \frac{t^k}{k!}$, $0\leq \alpha < 1$ both in the subcritical
case $\beta < \beta_N$ and critical case $\beta =\beta_N$ with $\beta_N = N
\omega_{N-1}^{\frac1{N-1}}$ and $\omega_{N-1}$ denotes the surface area of the
unit sphere in $\mathbb R^N$. We will show that $MT(N,\beta,\alpha)$ is
attained in the subcritical case if $N\geq 3$ or $N=2$ and $\beta \in
(\frac{2(1+2\alpha)}{(1+\alpha)^2 B_2},\beta_2)$ with $B_2$ is the best
constant in a Gagliardo--Nirenberg inequality in $W^{1,2}(\mathbb R^2)$. We
also show that $MT(2,\beta,\alpha)$ is not attained for $\beta$ small which is
different from the context of bounded domains. In the critical case, we prove
that $MT(N,\beta_N,\alpha)$ is attained for $\alpha\geq 0$ small enough. To
prove our results, we first establish a lower bound for $MT(N,\beta,\alpha)$
which excludes the concentrating or vanishing behaviors of their maximizer
sequences. This implies the attainability of $MT(N,\beta,\alpha)$ in the
subcritical case. The proof in the critical case is based on the blow-up
analysis method. Finally, by using the Moser sequence together the scaling
argument, we show that $MT(N,\beta_N,1) =\infty$. Our results settle the
questions left open in \cite{doO2015,doO2016}.
| 0 | 0 | 1 | 0 | 0 | 0 |
Dex: Incremental Learning for Complex Environments in Deep Reinforcement Learning | This paper introduces Dex, a reinforcement learning environment toolkit
specialized for training and evaluation of continual learning methods as well
as general reinforcement learning problems. We also present the novel continual
learning method of incremental learning, where a challenging environment is
solved using optimal weight initialization learned from first solving a similar
easier environment. We show that incremental learning can produce vastly
superior results than standard methods by providing a strong baseline method
across ten Dex environments. We finally develop a saliency method for
qualitative analysis of reinforcement learning, which shows the impact
incremental learning has on network attention.
| 1 | 0 | 0 | 1 | 0 | 0 |
The first order partial differential equations resolved with any derivatives | In this paper we discuss the first order partial differential equations
resolved with any derivatives. At first, we transform the first order partial
differential equation resolved with respect to a time derivative into a system
of linear equations. Secondly, we convert it into a system of the first order
linear partial differential equations with constant coefficients and nonlinear
algebraic equations. Thirdly, we solve them by the Fourier transform and
convert them into the equivalent integral equations. At last, we extend to
discuss the first order partial differential equations resolved with respect to
time derivatives and the general scenario resolved with any derivatives.
| 0 | 0 | 1 | 0 | 0 | 0 |
The Genus-One Global Mirror Theorem for the Quintic Threefold | We prove the genus-one restriction of the all-genus
Landau-Ginzburg/Calabi-Yau conjecture of Chiodo and Ruan, stated in terms of
the geometric quantization of an explicit symplectomorphism determined by
genus-zero invariants. This provides the first evidence supporting the
higher-genus Landau-Ginzburg/Calabi-Yau correspondence for the quintic
threefold, and exhibits the first instance of the "genus zero controls higher
genus" principle, in the sense of Givental's quantization formalism, for
non-semisimple cohomological field theories.
| 0 | 0 | 1 | 0 | 0 | 0 |
Self-Calibration of Mobile Manipulator Kinematic and Sensor Extrinsic Parameters Through Contact-Based Interaction | We present a novel approach for mobile manipulator self-calibration using
contact information. Our method, based on point cloud registration, is applied
to estimate the extrinsic transform between a fixed vision sensor mounted on a
mobile base and an end effector. Beyond sensor calibration, we demonstrate that
the method can be extended to include manipulator kinematic model parameters,
which involves a non-rigid registration process. Our procedure uses on-board
sensing exclusively and does not rely on any external measurement devices,
fiducial markers, or calibration rigs. Further, it is fully automatic in the
general case. We experimentally validate the proposed method on a custom mobile
manipulator platform, and demonstrate centimetre-level post-calibration
accuracy in positioning of the end effector using visual guidance only. We also
discuss the stability properties of the registration algorithm, in order to
determine the conditions under which calibration is possible.
| 1 | 0 | 0 | 0 | 0 | 0 |
Towards Open Data for the Citation Content Analysis | The paper presents first results of the CitEcCyr project funded by RANEPA.
The project aims to create a source of open citation data for research papers
written in Russian. Compared to existing sources of citation data, CitEcCyr is
working to provide the following added values: a) a transparent and distributed
architecture of a technology that generates the citation data; b) an openness
of all built/used software and created citation data; c) an extended set of
citation data sufficient for the citation content analysis; d) services for
public control over a quality of the citation data and a citing activity of
researchers.
| 1 | 0 | 0 | 0 | 0 | 0 |
Real Time Collision Detection and Identification for Robotic Manipulators | The majority of everyday tasks involve interacting with unstructured
environments. This implies that, in order for robots to be truly useful they
must be able to handle contacts. This paper explores how a particle filter can
be used to localize a contact point and estimate the external force. We
demonstrate the capability of the particle filter on a simulated 4DoF planar
robotic arm, and compare it to a well-established analytical approach.
| 1 | 0 | 0 | 0 | 0 | 0 |
Time-dynamic inference for non-Markov transition probabilities under independent right-censoring | In this article, weak convergence of the general non-Markov state transition
probability estimator by Titman (2015) is established which, up to now, has not
been verified yet for other general non-Markov estimators. A similar theorem is
shown for the bootstrap, yielding resampling-based inference methods for
statistical functionals. Formulas of the involved covariance functions are
presented in detail. Particular applications include the conditional expected
length of stay in a specific state, given occupation of another state in the
past, as well as the construction of time-simultaneous confidence bands for the
transition probabilities. The expected lengths of stay in the two-sample liver
cirrhosis data-set by Andersen et al. (1993) are compared and confidence
intervals for their difference are constructed. With borderline significance
and in comparison to the placebo group, the treatment group has an elevated
expected length of stay in the healthy state given an earlier disease state
occupation. In contrast, the Aalen-Johansen estimator-based confidence
interval, which relies on a Markov assumption, leads to a drastically different
conclusion. Also, graphical illustrations of confidence bands for the
transition probabilities demonstrate the biasedness of the Aalen-Johansen
estimator in this data example. The reliability of these results is assessed in
a simulation study.
| 0 | 0 | 1 | 1 | 0 | 0 |
Multitask Learning with CTC and Segmental CRF for Speech Recognition | Segmental conditional random fields (SCRFs) and connectionist temporal
classification (CTC) are two sequence labeling methods used for end-to-end
training of speech recognition models. Both models define a transcription
probability by marginalizing decisions about latent segmentation alternatives
to derive a sequence probability: the former uses a globally normalized joint
model of segment labels and durations, and the latter classifies each frame as
either an output symbol or a "continuation" of the previous label. In this
paper, we train a recognition model by optimizing an interpolation between the
SCRF and CTC losses, where the same recurrent neural network (RNN) encoder is
used for feature extraction for both outputs. We find that this multitask
objective improves recognition accuracy when decoding with either the SCRF or
CTC models. Additionally, we show that CTC can also be used to pretrain the RNN
encoder, which improves the convergence rate when learning the joint model.
| 1 | 0 | 0 | 0 | 0 | 0 |
Electronic structure and non-linear optical properties of organic photovoltaic systems with potential applications on solar cell devices: A DFT approach | The use of eco-friendly materials for the environment has been addressed as a
critical issue in the development of systems for renewable energy applications.
In this regard, the investigation of organic photovoltaic (OPV) molecules for
the implementation in solar cells, has become a subject of intense research in
the last years. The present work is a systematic study at the B3LYP level of
theory performed for a series of 50 OPV materials. Full geometry optimizations
revealed that those systems with a twisted geometry are the most energetically
stable. Nuclear independent Chemical shifts (NICS) values show a strong
aromatic character along the series, indicating a possible polymerization in
solid-state, via a {\pi}-{\pi} stacking, which may be relevant in the design of
a solar cell device. The absorption spectra in the series was also computed
using Time Dependent DFT at the same level of theory, indicating that all
spectra are red-shifted along the series. This is a promissory property that
may be directly implemented in a photovoltaic material, since it is possible to
absorb a larger range of visible light. The computed HOMO-LUMO gaps as a
measurement of the band gap in semiconductors, show a reasonable agreement with
those found in experiment, predicting candidate materials that may be directly
used in photovoltaic applications. Non-linear optical (NLO) properties were
also estimated with the aid of a PCBM molecule as a model of an acceptor, and a
final set of optimal systems was identified as potential candidates to be
implemented as photovoltaic materials. The methodological approach presented in
this work may aid in the in silico assisted-design of OPV materials.
| 0 | 1 | 0 | 0 | 0 | 0 |
Connectivity Learning in Multi-Branch Networks | While much of the work in the design of convolutional networks over the last
five years has revolved around the empirical investigation of the importance of
depth, filter sizes, and number of feature channels, recent studies have shown
that branching, i.e., splitting the computation along parallel but distinct
threads and then aggregating their outputs, represents a new promising
dimension for significant improvements in performance. To combat the complexity
of design choices in multi-branch architectures, prior work has adopted simple
strategies, such as a fixed branching factor, the same input being fed to all
parallel branches, and an additive combination of the outputs produced by all
branches at aggregation points.
In this work we remove these predefined choices and propose an algorithm to
learn the connections between branches in the network. Instead of being chosen
a priori by the human designer, the multi-branch connectivity is learned
simultaneously with the weights of the network by optimizing a single loss
function defined with respect to the end task. We demonstrate our approach on
the problem of multi-class image classification using three different datasets
where it yields consistently higher accuracy compared to the state-of-the-art
"ResNeXt" multi-branch network given the same learning capacity.
| 1 | 0 | 0 | 0 | 0 | 0 |
A multiple timescales approach to bridging spiking- and population-level dynamics | A rigorous bridge between spiking-level and macroscopic quantities is an
on-going and well-developed story for asynchronously firing neurons, but focus
has shifted to include neural populations exhibiting varying synchronous
dynamics. Recent literature has used the Ott--Antonsen ansatz (2008) to great
effect, allowing a rigorous derivation of an order parameter for large
oscillator populations. The ansatz has been successfully applied using several
models including networks of Kuramoto oscillators, theta models, and
integrate-and-fire neurons, along with many types of network topologies. In the
present study, we take a converse approach: given the mean field dynamics of
slow synapses, predict the synchronization properties of finite neural
populations. The slow synapse assumption is amenable to averaging theory and
the method of multiple timescales. Our proposed theory applies to two
heterogeneous populations of N excitatory n-dimensional and N inhibitory
m-dimensional oscillators with homogeneous synaptic weights. We then
demonstrate our theory using two examples. In the first example we take a
network of excitatory and inhibitory theta neurons and consider the case with
and without heterogeneous inputs. In the second example we use Traub models
with calcium for the excitatory neurons and Wang-Buzs{á}ki models for the
inhibitory neurons. We accurately predict phase drift and phase locking in each
example even when the slow synapses exhibit non-trivial mean-field dynamics.
| 0 | 0 | 0 | 0 | 1 | 0 |
Two-Stream RNN/CNN for Action Recognition in 3D Videos | The recognition of actions from video sequences has many applications in
health monitoring, assisted living, surveillance, and smart homes. Despite
advances in sensing, in particular related to 3D video, the methodologies to
process the data are still subject to research. We demonstrate superior results
by a system which combines recurrent neural networks with convolutional neural
networks in a voting approach. The gated-recurrent-unit-based neural networks
are particularly well-suited to distinguish actions based on long-term
information from optical tracking data; the 3D-CNNs focus more on detailed,
recent information from video data. The resulting features are merged in an SVM
which then classifies the movement. In this architecture, our method improves
recognition rates of state-of-the-art methods by 14% on standard data sets.
| 1 | 0 | 0 | 0 | 0 | 0 |
DNA translocation through alpha-haemolysin nano-pores with potential application to macromolecular data storage | Digital information can be encoded in the building-block sequence of
macromolecules, such as RNA and single-stranded DNA. Methods of "writing" and
"reading" macromolecular strands are currently available, but they are slow and
expensive. In an ideal molecular data storage system, routine operations such
as write, read, erase, store, and transfer must be done reliably and at high
speed within an integrated chip. As a first step toward demonstrating the
feasibility of this concept, we report preliminary results of DNA readout
experiments conducted in miniaturized chambers that are scalable to even
smaller dimensions. We show that translocation of a single-stranded DNA
molecule (consisting of 50 adenosine bases followed by 100 cytosine bases)
through an ion-channel yields a characteristic signal that is attributable to
the 2-segment structure of the molecule. We also examine the dependence of the
translocation rate and speed on the adjustable parameters of the experiment.
| 1 | 1 | 0 | 0 | 0 | 0 |
Contextuality from missing and versioned data | Traditionally categorical data analysis (e.g. generalized linear models)
works with simple, flat datasets akin to a single table in a database with no
notion of missing data or conflicting versions. In contrast, modern data
analysis must deal with distributed databases with many partial local tables
that need not always agree. The computational agents tabulating these tables
are spatially separated, with binding speed-of-light constraints and data
arriving too rapidly for these distributed views ever to be fully informed and
globally consistent. Contextuality is a mathematical property which describes a
kind of inconsistency arising in quantum mechanics (e.g. in Bell's theorem). In
this paper we show how contextuality can arise in common data collection
scenarios, including missing data and versioning (as in low-latency distributed
databases employing snapshot isolation). In the companion paper, we develop
statistical models adapted to this regime.
| 1 | 0 | 0 | 1 | 0 | 0 |
Reveal the Mantle and K-40 Components of Geoneutrinos with Liquid Scintillator Cherenkov Neutrino Detectors | In this article we present an idea of using liquid scintillator Cherenkov
neutrino detectors to detect the mantle and K-40 components of geoneutrinos.
Liquid scintillator Cherenkov detectors feature both energy and direction
measurement for charge particles. Geoneutrinos can be detected with the elastic
scattering process of neutrino and electron. With the directionality, the
dominant intrinsic background originated from solar neutrinos in common liquid
scintillator detectors can be suppressed. The mantle geoneutrinos can be
distinguished because they come mainly underneath. The K-40 geoneutrinos can
also be identified, if the detection threshold for direction measurement can be
lower than, for example, 0.8 MeV. According to our calculation, a moderate,
kilo-ton scale, detector can observe tens of candidates, and is a practical
start for an experiment.
| 0 | 1 | 0 | 0 | 0 | 0 |
Improving Stock Movement Prediction with Adversarial Training | This paper contributes a new machine learning solution for stock movement
prediction, which aims to predict whether the price of a stock will be up or
down in the near future. The key novelty is that we propose to employ
adversarial training to improve the generalization of a recurrent neural
network model. The rationality of adversarial training here is that the input
features to stock prediction are typically based on stock price, which is
essentially a stochastic variable and continuously changed with time by nature.
As such, normal training with stationary price-based features (e.g. the closing
price) can easily overfit the data, being insufficient to obtain reliable
models. To address this problem, we propose to add perturbations to simulate
the stochasticity of continuous price variable, and train the model to work
well under small yet intentional perturbations. Extensive experiments on two
real-world stock data show that our method outperforms the state-of-the-art
solution with 3.11% relative improvements on average w.r.t. accuracy, verifying
the usefulness of adversarial training for stock prediction task. Codes will be
made available upon acceptance.
| 0 | 0 | 0 | 0 | 0 | 1 |
Weighted estimates for positive operators and Doob maximal operators on filtered measure spaces | We characterize strong type and weak type inequalities with two weights for
positive operators on filtered measure spaces. These estimates are
probabilistic analogues of two-weight inequalities for positive operators
associated to the dyadic cubes in $\mathbb R^n$ due to Lacey, Sawyer and
Uriarte-Tuero \cite{LaSaUr}. Several mixed bounds for the Doob maximal operator
on filtered measure spaces are also obtained. In fact, Hytönen-Pérez
type and Lerner-Moen type norm estimates for Doob maximal operator are
established. Our approaches are mainly based on the construction of principal
sets.
| 0 | 0 | 1 | 0 | 0 | 0 |
Sentiment Identification in Code-Mixed Social Media Text | Sentiment analysis is the Natural Language Processing (NLP) task dealing with
the detection and classification of sentiments in texts. While some tasks deal
with identifying the presence of sentiment in the text (Subjectivity analysis),
other tasks aim at determining the polarity of the text categorizing them as
positive, negative and neutral. Whenever there is a presence of sentiment in
the text, it has a source (people, group of people or any entity) and the
sentiment is directed towards some entity, object, event or person. Sentiment
analysis tasks aim to determine the subject, the target and the polarity or
valence of the sentiment. In our work, we try to automatically extract
sentiment (positive or negative) from Facebook posts using a machine learning
approach.While some works have been done in code-mixed social media data and in
sentiment analysis separately, our work is the first attempt (as of now) which
aims at performing sentiment analysis of code-mixed social media text. We have
used extensive pre-processing to remove noise from raw text. Multilayer
Perceptron model has been used to determine the polarity of the sentiment. We
have also developed the corpus for this task by manually labeling Facebook
posts with their associated sentiments.
| 1 | 0 | 0 | 0 | 0 | 0 |
SVSGAN: Singing Voice Separation via Generative Adversarial Network | Separating two sources from an audio mixture is an important task with many
applications. It is a challenging problem since only one signal channel is
available for analysis. In this paper, we propose a novel framework for singing
voice separation using the generative adversarial network (GAN) with a
time-frequency masking function. The mixture spectra is considered to be a
distribution and is mapped to the clean spectra which is also considered a
distribtution. The approximation of distributions between mixture spectra and
clean spectra is performed during the adversarial training process. In contrast
with current deep learning approaches for source separation, the parameters of
the proposed framework are first initialized in a supervised setting and then
optimized by the training procedure of GAN in an unsupervised setting.
Experimental results on three datasets (MIR-1K, iKala and DSD100) show that
performance can be improved by the proposed framework consisting of
conventional networks.
| 1 | 0 | 0 | 0 | 0 | 0 |
Microscopic theory of refractive index applied to metamaterials: Effective current response tensor corresponding to standard relation $n^2 = \varepsilon_{\mathrm{eff}} μ_{\mathrm{eff}}$ | In this article, we first derive the wavevector- and frequency-dependent,
microscopic current response tensor which corresponds to the "macroscopic"
ansatz $\vec D = \varepsilon_0 \varepsilon_{\mathrm{eff}} \vec E$ and $\vec B =
\mu_0 \mu_{\mathrm{eff}} \vec H$ with wavevector- and frequency-independent,
"effective" material constants $\varepsilon_{\mathrm{eff}}$ and
$\mu_{\mathrm{eff}}$. We then deduce the electromagnetic and optical properties
of this effective material model by employing exact, microscopic response
relations. In particular, we argue that for recovering the standard relation
$n^2 = \varepsilon_{\mathrm{eff}} \mu_{\mathrm{eff}}$ between the refractive
index and the effective material constants, it is imperative to start from the
microscopic wave equation in terms of the transverse dielectric function,
$\varepsilon_{\mathrm{T}}(\vec k, \omega) = 0$. On the phenomenological side,
our result is especially relevant for metamaterials research, which draws
directly on the standard relation for the refractive index in terms of
effective material constants. Since for a wide class of materials the current
response tensor can be calculated from first principles and compared to the
model expression derived here, this work also paves the way for a systematic
search for new metamaterials.
| 0 | 1 | 0 | 0 | 0 | 0 |
Semidefinite Relaxation-Based Optimization of Multiple-Input Wireless Power Transfer Systems | An optimization procedure for multi-transmitter (MISO) wireless power
transfer (WPT) systems based on tight semidefinite relaxation (SDR) is
presented. This method ensures physical realizability of MISO WPT systems
designed via convex optimization -- a robust, semi-analytical and intuitive
route to optimizing such systems. To that end, the nonconvex constraints
requiring that power is fed into rather than drawn from the system via all
transmitter ports are incorporated in a convex semidefinite relaxation, which
is efficiently and reliably solvable by dedicated algorithms. A test of the
solution then confirms that this modified problem is equivalent (tight
relaxation) to the original (nonconvex) one and that the true global optimum
has been found. This is a clear advantage over global optimization methods
(e.g. genetic algorithms), where convergence to the true global optimum cannot
be ensured or tested. Discussions of numerical results yielded by both the
closed-form expressions and the refined technique illustrate the importance and
practicability of the new method. It, is shown that this technique offers a
rigorous optimization framework for a broad range of current and emerging WPT
applications.
| 1 | 0 | 1 | 0 | 0 | 0 |
Counterfactual Reasoning with Disjunctive Knowledge in a Linear Structural Equation Model | We consider the problem of estimating counterfactual quantities when prior
knowledge is available in the form of disjunctive statements. These include
disjunction of conditions (e.g., "the patient is more than 60 years of age") as
well as disjuction of antecedants (e.g., "had the patient taken either drug A
or drug B"). Focusing on linear structural equation models (SEM) and imperfect
control plans, we extend the counterfactual framework of Balke and Pearl (1995)
, Chen and Pearl (2015), and Pearl (2009, pp. 389-391) from unconditional to
conditional plans, from a univariate treatment to a set of treatments, and from
point type knowledge to disjunctive knowledge. Finally, we provide improved
matrix representations of the resulting counterfactual parameters, and improved
computational methods of their evaluation.
| 0 | 0 | 0 | 1 | 0 | 0 |
Post-edit Analysis of Collective Biography Generation | Text generation is increasingly common but often requires manual post-editing
where high precision is critical to end users. However, manual editing is
expensive so we want to ensure this effort is focused on high-value tasks. And
we want to maintain stylistic consistency, a particular challenge in crowd
settings. We present a case study, analysing human post-editing in the context
of a template-based biography generation system. An edit flow visualisation
combined with manual characterisation of edits helps identify and prioritise
work for improving end-to-end efficiency and accuracy.
| 1 | 0 | 0 | 0 | 0 | 0 |
The SysML/KAOS Domain Modeling Approach | A means of building safe critical systems consists of formally modeling the
requirements formulated by stakeholders and ensuring their consistency with
respect to application domain properties. This paper proposes a metamodel for
an ontology modeling formalism based on OWL and PLIB. This modeling formalism
is part of a method for modeling the domain of systems whose requirements are
captured through SysML/KAOS. The formal semantics of SysML/KAOS goals are
represented using Event-B specifications. Goals provide the set of events,
while domain models will provide the structure of the system state of the
Event-B specification. Our proposal is illustrated through a case study dealing
with a Cycab localization component specification. The case study deals with
the specification of a localization software component that uses GPS,Wi-Fi and
sensor technologies for the realtime localization of the Cycab vehicle, an
autonomous ground transportation system designed to be robust and completely
independent.
| 1 | 0 | 0 | 0 | 0 | 0 |
Neighborhood selection with application to social networks | The topic of this paper is modeling and analyzing dependence in stochastic
social networks. Using a latent variable block model allows the analysis of
dependence between blocks via the analysis of a latent graphical model. Our
approach to the analysis of the graphical model then is based on the idea
underlying the neighborhood selection scheme put forward by Meinshausen and
Bühlmann (2006). However, because of the latent nature of our model,
estimates have to be used in lieu of the unobserved variables. This leads to a
novel analysis of graphical models under uncertainty, in the spirit of
Rosenbaum et al. (2010), or Belloni et al. (2017). Lasso-based selectors, and a
class of Dantzig-type selectors are studied.
| 0 | 0 | 1 | 1 | 0 | 0 |
Spinless hourglass nodal-line semimetals | Nodal-line semimetals, one of the topological semimetals, have degeneracy
along nodal lines where the band gap is closed. In many cases, the nodal lines
appear accidentally, and in such cases it is impossible to determine whether
the nodal lines appear or not, only from the crystal symmetry and the electron
filling. In this paper, for spinless systems, we show that in specific space
groups at $4N+2$ fillings ($8N+4$ fillings including the spin degree of
freedom), presence of the nodal lines is required regardless of the details of
the systems. Here, the spinless systems refer to crystals where the spin-orbit
coupling is negligible and the spin degree of freedom can be omitted because of
the SU(2) spin degeneracy. In this case the shape of the band structure around
these nodal lines is like an hourglass, and we call this a spinless hourglass
nodal-line semimetal. We construct a model Hamiltonian as an example and we
show that it is always in the spinless hourglass nodal-line semimetal phase
even when the model parameters are changed without changing the symmetries of
the system. We also establish a list of all the centrosymmetric space groups,
under which spinless systems always have hourglass nodal lines, and illustrate
where the nodal lines are located. We propose that Al$_3$FeSi$_2$, whose
space-group symmetry is Pbcn (No. 60), is one of the nodal-line semimetals
arising from this mechanism.
| 0 | 1 | 0 | 0 | 0 | 0 |
Average-radius list-recovery of random linear codes: it really ties the room together | We analyze the list-decodability, and related notions, of random linear
codes. This has been studied extensively before: there are many different
parameter regimes and many different variants. Previous works have used
complementary styles of arguments---which each work in their own parameter
regimes but not in others---and moreover have left some gaps in our
understanding of the list-decodability of random linear codes. In particular,
none of these arguments work well for list-recovery, a generalization of
list-decoding that has been useful in a variety of settings.
In this work, we present a new approach, which works across parameter regimes
and further generalizes to list-recovery. Our main theorem can establish better
list-decoding and list-recovery results for low-rate random linear codes over
large fields; list-recovery of high-rate random linear codes; and it can
recover the rate bounds of Guruswami, Hastad, and Kopparty for constant-rate
random linear codes (although with large list sizes).
| 1 | 0 | 0 | 0 | 0 | 0 |
Insights on representational similarity in neural networks with canonical correlation | Comparing different neural network representations and determining how
representations evolve over time remain challenging open questions in our
understanding of the function of neural networks. Comparing representations in
neural networks is fundamentally difficult as the structure of representations
varies greatly, even across groups of networks trained on identical tasks, and
over the course of training. Here, we develop projection weighted CCA
(Canonical Correlation Analysis) as a tool for understanding neural networks,
building off of SVCCA, a recently proposed method (Raghu et al., 2017). We
first improve the core method, showing how to differentiate between signal and
noise, and then apply this technique to compare across a group of CNNs,
demonstrating that networks which generalize converge to more similar
representations than networks which memorize, that wider networks converge to
more similar solutions than narrow networks, and that trained networks with
identical topology but different learning rates converge to distinct clusters
with diverse representations. We also investigate the representational dynamics
of RNNs, across both training and sequential timesteps, finding that RNNs
converge in a bottom-up pattern over the course of training and that the hidden
state is highly variable over the course of a sequence, even when accounting
for linear transforms. Together, these results provide new insights into the
function of CNNs and RNNs, and demonstrate the utility of using CCA to
understand representations.
| 0 | 0 | 0 | 1 | 0 | 0 |
Complex Hadamard matrices with noncommutative entries | We axiomatize and study the matrices of type $H\in M_N(A)$, having unitary
entries, $H_{ij}\in U(A)$, and whose rows and columns are subject to
orthogonality type conditions. Here $A$ can be any $C^*$-algebra, for instance
$A=\mathbb C$, where we obtain the usual complex Hadamard matrices, or
$A=C(X)$, where we obtain the continuous families of complex Hadamard matrices.
Our formalism allows the construction of a quantum permutation group $G\subset
S_N^+$, whose structure and computation is discussed here.
| 0 | 0 | 1 | 0 | 0 | 0 |
Subregular Complexity and Deep Learning | This paper argues that the judicial use of formal language theory and
grammatical inference are invaluable tools in understanding how deep neural
networks can and cannot represent and learn long-term dependencies in temporal
sequences. Learning experiments were conducted with two types of Recurrent
Neural Networks (RNNs) on six formal languages drawn from the Strictly Local
(SL) and Strictly Piecewise (SP) classes. The networks were Simple RNNs
(s-RNNs) and Long Short-Term Memory RNNs (LSTMs) of varying sizes. The SL and
SP classes are among the simplest in a mathematically well-understood hierarchy
of subregular classes. They encode local and long-term dependencies,
respectively. The grammatical inference algorithm Regular Positive and Negative
Inference (RPNI) provided a baseline. According to earlier research, the LSTM
architecture should be capable of learning long-term dependencies and should
outperform s-RNNs. The results of these experiments challenge this narrative.
First, the LSTMs' performance was generally worse in the SP experiments than in
the SL ones. Second, the s-RNNs out-performed the LSTMs on the most complex SP
experiment and performed comparably to them on the others.
| 1 | 0 | 0 | 0 | 0 | 0 |
Observation of oscillatory relaxation in the Sn-terminated surface of epitaxial rock-salt SnSe $\{111\}$ topological crystalline insulator | Topological crystalline insulators have been recently predicted and observed
in rock-salt structure SnSe $\{111\}$ thin films. Previous studies have
suggested that the Se-terminated surface of this thin film with hydrogen
passivation, has a reduced surface energy and is thus a preferred
configuration. In this paper, synchrotron-based angle-resolved photoemission
spectroscopy, along with density functional theory calculations, are used to
demonstrate conclusively that a rock-salt SnSe $\{111\}$ thin film
epitaxially-grown on \ce{Bi2Se3} has a stable Sn-terminated surface. These
observations are supported by low energy electron diffraction (LEED)
intensity-voltage measurements and dynamical LEED calculations, which further
show that the Sn-terminated SnSe $\{111\}$ thin film has undergone a surface
structural relaxation of the interlayer spacing between the Sn and Se atomic
planes. In sharp contrast to the Se-terminated counterpart, the observed Dirac
surface state in the Sn-terminated SnSe $\{111\}$ thin film is shown to yield a
high Fermi velocity, $0.50\times10^6$m/s, which suggests a potential mechanism
of engineering the Dirac surface state of topological materials by tuning the
surface configuration.
| 0 | 1 | 0 | 0 | 0 | 0 |
Chaotic Dynamics of Inner Ear Hair Cells | Experimental records of active bundle motility are used to demonstrate the
presence of a low-dimensional chaotic attractor in hair cell dynamics.
Dimensionality tests from dynamic systems theory are applied to estimate the
number of independent variables sufficient for modeling the hair cell response.
Poincare maps are constructed to observe a quasiperiodic transition from chaos
to order with increasing amplitudes of mechanical forcing. The onset of this
transition is accompanied by a reduction of Kolmogorov entropy in the system
and an increase in mutual information between the stimulus and the hair bundle,
indicative of signal detection. A simple theoretical model is used to describe
the observed chaotic dynamics. The model exhibits an enhancement of sensitivity
to weak stimuli when the system is poised in the chaotic regime. We propose
that chaos may play a role in the hair cell's ability to detect low-amplitude
sounds.
| 0 | 1 | 0 | 0 | 0 | 0 |
From Dirac semimetals to topological phases in three dimensions: a coupled wire construction | Weyl and Dirac (semi)metals in three dimensions have robust gapless
electronic band structures. Their massless single-body energy spectra are
protected by symmetries such as lattice translation, (screw) rotation and time
reversal. In this manuscript, we discuss many-body interactions in these
systems. We focus on strong interactions that preserve symmetries and are
outside the single-body mean-field regime. By mapping a Dirac (semi)metal to a
model based on a three dimensional array of coupled Dirac wires, we show (1)
the Dirac (semi)metal can acquire a many-body excitation energy gap without
breaking the relevant symmetries, and (2) interaction can enable an anomalous
Weyl (semi)metallic phase that is otherwise forbidden by symmetries in the
single-body setting and can only be present holographically on the boundary of
a four dimensional weak topological insulator. Both of these topological states
support fractional gapped (gapless) bulk (resp. boundary) quasiparticle
excitations.
| 0 | 1 | 0 | 0 | 0 | 0 |
CHIME FRB: An application of FFT beamforming for a radio telescope | We have developed FFT beamforming techniques for the CHIME radio telescope,
to search for and localize the astrophysical signals from Fast Radio Bursts
(FRBs) over a large instantaneous field-of-view (FOV) while maintaining the
full angular resolution of CHIME. We implement a hybrid beamforming pipeline in
a GPU correlator, synthesizing 256 FFT-formed beams in the North-South
direction by four formed beams along East-West via exact phasing, tiling a sky
area of ~250 square degrees. A zero-padding approximation is employed to
improve chromatic beam alignment across the wide bandwidth of 400 to 800 MHz.
We up-channelize the data in order to achieve fine spectral resolution of
$\Delta\nu$=24 kHz and time cadence of 0.983 ms, desirable for detecting
transient and dispersed signals such as those from FRBs.
| 0 | 1 | 0 | 0 | 0 | 0 |
MRI-PET Registration with Automated Algorithm in Pre-clinical Studies | Magnetic Resonance Imaging (MRI) and Positron Emission Tomography (PET)
automatic 3-D registration is implemented and validated for small animal image
volumes so that the high-resolution anatomical MRI information can be fused
with the low spatial resolution of functional PET information for the
localization of lesion that is currently in high demand in the study of tumor
of cancer (oncology) and its corresponding preparation of pharmaceutical drugs.
Though many registration algorithms are developed and applied on human brain
volumes, these methods may not be as efficient on small animal datasets due to
lack of intensity information and often the high anisotropy in voxel
dimensions. Therefore, a fully automatic registration algorithm which can
register not only assumably rigid small animal volumes such as brain but also
deformable organs such as kidney, cardiac and chest is developed using a
combination of global affine and local B-spline transformation models in which
mutual information is used as a similarity criterion. The global affine
registration uses a multi-resolution pyramid on image volumes of 3 levels
whereas in local B-spline registration, a multi-resolution scheme is applied on
the B-spline grid of 2 levels on the finest resolution of the image volumes in
which only the transform itself is affected rather than the image volumes.
Since mutual information lacks sufficient spatial information, PCA is used to
inject it by estimating initial translation and rotation parameters. It is
computationally efficient since it is implemented using C++ and ITK library,
and is qualitatively and quantitatively shown that this PCA-initialized global
registration followed by local registration is in close agreement with expert
manual registration and outperforms the one without PCA initialization tested
on small animal brain and kidney.
| 1 | 0 | 0 | 0 | 0 | 0 |
Generalization of Special Functions and its Applications to Multiplicative and Ordinary Fractional Derivatives | The goal of this paper is to extend the classical and multiplicative
fractional derivatives. For this purpose, it is introduced the new extended
modified Bessel function and also given an important relation between this new
function I(v,q;x) and the confluent hypergeometric function. Besides, it is
used to generalize the hypergeometric, the confluent hypergeometric and the
extended beta functions by using the new extended modified Bessel function.
Also, the asymptotic formulae and the generating function of the extended
modified Bessel function are obtained. The extensions of classical and
multiplicative fractional derivatives are defined via extended modified Bessel
function and, first time the fractional derivative of rational functions is
explicitly given via complex partial fraction decomposition.
| 0 | 0 | 1 | 0 | 0 | 0 |
Incident Light Frequency-based Image Defogging Algorithm | Considering the problem of color distortion caused by the defogging algorithm
based on dark channel prior, an improved algorithm was proposed to calculate
the transmittance of all channels respectively. First, incident light
frequency's effect on the transmittance of various color channels was analyzed
according to the Beer-Lambert's Law, from which a proportion among various
channel transmittances was derived; afterwards, images were preprocessed by
down-sampling to refine transmittance, and then the original size was restored
to enhance the operational efficiency of the algorithm; finally, the
transmittance of all color channels was acquired in accordance with the
proportion, and then the corresponding transmittance was used for image
restoration in each channel. The experimental results show that compared with
the existing algorithm, this improved image defogging algorithm could make
image colors more natural, solve the problem of slightly higher color
saturation caused by the existing algorithm, and shorten the operation time by
four to nine times.
| 1 | 0 | 0 | 0 | 0 | 0 |
Tangent points of d-lower content regular sets and $β$ numbers | We present a generalisation of C. Bishop and P. Jones' result in [BJ1], where
they give a characterisation of the tangent points of a Jordan curve in terms
of $\beta$ numbers. Instead of the $L^\infty$ Jones' $\beta$ numbers, we use an
averaged version of them, firstly introduced by J. Azzam and R. Schul in [AS1].
A fundamental tool in the proof will be the Reifenberg parameterisation Theorem
of G. David and T. Toro (see [DT1]).
| 0 | 0 | 1 | 0 | 0 | 0 |
Quasi-Oracle Estimation of Heterogeneous Treatment Effects | Flexible estimation of heterogeneous treatment effects lies at the heart of
many statistical challenges, such as personalized medicine and optimal resource
allocation. In this paper, we develop a general class of two-step algorithms
for heterogeneous treatment effect estimation in observational studies. We
first estimate marginal effects and treatment propensities in order to form an
objective function that isolates the causal component of the signal. Then, we
optimize this data-adaptive objective function. Our approach has several
advantages over existing methods. From a practical perspective, our method is
flexible and easy to use: In both steps, we can use any loss-minimization
method, e.g., penalized regression, deep neutral networks, or boosting;
moreover, these methods can be fine-tuned by cross validation. Meanwhile, in
the case of penalized kernel regression, we show that our method has a
quasi-oracle property: Even if the pilot estimates for marginal effects and
treatment propensities are not particularly accurate, we achieve the same error
bounds as an oracle who has a priori knowledge of these two nuisance
components. We implement variants of our approach based on both penalized
regression and boosting in a variety of simulation setups, and find promising
performance relative to existing baselines.
| 0 | 0 | 1 | 1 | 0 | 0 |
Adaptive clustering procedure for continuous gravitational wave searches | In hierarchical searches for continuous gravitational waves, clustering of
candidates is an important postprocessing step because it reduces the number of
noise candidates that are followed-up at successive stages [1][7][12]. Previous
clustering procedures bundled together nearby candidates ascribing them to the
same root cause (be it a signal or a disturbance), based on a predefined
cluster volume. In this paper, we present a procedure that adapts the cluster
volume to the data itself and checks for consistency of such volume with what
is expected from a signal. This significantly improves the noise rejection
capabilities at fixed detection threshold, and at fixed computing resources for
the follow-up stages, this results in an overall more sensitive search. This
new procedure was employed in the first Einstein@Home search on data from the
first science run of the advanced LIGO detectors (O1) [11].
| 0 | 1 | 1 | 0 | 0 | 0 |
The heat trace for the drifting Laplacian and Schrödinger operators on manifolds | We study the heat trace for both the drifting Laplacian as well as
Schrödinger operators on compact Riemannian manifolds. In the case of a
finite regularity potential or weight function, we prove the existence of a
partial (six term) asymptotic expansion of the heat trace for small times as
well as a suitable remainder estimate. We also demonstrate that the more
precise asymptotic behavior of the remainder is determined by and conversely
distinguishes higher (Sobolev) regularity on the potential or weight function.
In the case of a smooth weight function, we determine the full asymptotic
expansion of the heat trace for the drifting Laplacian for small times. We then
use the heat trace to study the asymptotics of the eigenvalue counting
function. In both cases the Weyl law coincides with the Weyl law for the
Riemannian manifold with the standard Laplace-Beltrami operator. We conclude by
demonstrating isospectrality results for the drifting Laplacian on compact
manifolds.
| 0 | 0 | 1 | 0 | 0 | 0 |
Generative Adversarial Networks for Black-Box API Attacks with Limited Training Data | As online systems based on machine learning are offered to public or paid
subscribers via application programming interfaces (APIs), they become
vulnerable to frequent exploits and attacks. This paper studies adversarial
machine learning in the practical case when there are rate limitations on API
calls. The adversary launches an exploratory (inference) attack by querying the
API of an online machine learning system (in particular, a classifier) with
input data samples, collecting returned labels to build up the training data,
and training an adversarial classifier that is functionally equivalent and
statistically close to the target classifier. The exploratory attack with
limited training data is shown to fail to reliably infer the target classifier
of a real text classifier API that is available online to the public. In
return, a generative adversarial network (GAN) based on deep learning is built
to generate synthetic training data from a limited number of real training data
samples, thereby extending the training data and improving the performance of
the inferred classifier. The exploratory attack provides the basis to launch
the causative attack (that aims to poison the training process) and evasion
attack (that aims to fool the classifier into making wrong decisions) by
selecting training and test data samples, respectively, based on the confidence
scores obtained from the inferred classifier. These stealth attacks with small
footprint (using a small number of API calls) make adversarial machine learning
practical under the realistic case with limited training data available to the
adversary.
| 1 | 0 | 0 | 1 | 0 | 0 |
Coloring ($P_6$, diamond, $K_4$)-free graphs | We show that every ($P_6$, diamond, $K_4$)-free graph is $6$-colorable.
Moreover, we give an example of a ($P_6$, diamond, $K_4$)-free graph $G$ with
$\chi(G) = 6$. This generalizes some known results in the literature.
| 1 | 0 | 0 | 0 | 0 | 0 |
Habitable Climate Scenarios for Proxima Centauri b With a Dynamic Ocean | The nearby exoplanet Proxima Centauri b will be a prime future target for
characterization, despite questions about its retention of water. Climate
models with static oceans suggest that an Earth-like Proxima b could harbor a
small dayside region of surface liquid water at fairly warm temperatures
despite its weak instellation. We present the first 3-dimensional climate
simulations of Proxima b with a dynamic ocean. We find that an ocean-covered
Proxima b could have a much broader area of surface liquid water but at much
colder temperatures than previously suggested, due to ocean heat transport and
depression of the freezing point by salinity. Elevated greenhouse gas
concentrations do not necessarily produce more open ocean area because of
possible dynamic regime transitions. For an evolutionary path leading to a
highly saline present ocean, Proxima b could conceivably be an inhabited,
mostly open ocean planet dominated by halophilic life. For an ocean planet in
3:2 spin-orbit resonance, a permanent tropical waterbelt exists for moderate
eccentricity. Simulations of Proxima Centauri b may also be a model for the
habitability of planets receiving similar instellation from slightly cooler or
warmer stars, e.g., in the TRAPPIST-1, LHS 1140, GJ 273, and GJ 3293 systems.
| 0 | 1 | 0 | 0 | 0 | 0 |
Level bounds for exceptional quantum subgroups in rank two | There is a long-standing belief that the modular tensor categories
$\mathcal{C}(\mathfrak{g},k)$, for $k\in\mathbb{Z}_{\geq1}$ and
finite-dimensional simple complex Lie algebras $\mathfrak{g}$, contain
exceptional connected étale algebras at only finitely many levels $k$. This
premise has known implications for the study of relations in the Witt group of
nondegenerate braided fusion categories, modular invariants of conformal field
theories, and the classification of subfactors in the theory of von Neumann
algebras. Here we confirm this conjecture when $\mathfrak{g}$ has rank 2,
contributing proofs and explicit bounds when $\mathfrak{g}$ is of type $B_2$ or
$G_2$, adding to the previously known positive results for types $A_1$ and
$A_2$.
| 0 | 0 | 1 | 0 | 0 | 0 |
Convolution Forgetting Curve Model for Repeated Learning | Most of mathematic forgetting curve models fit well with the forgetting data
under the learning condition of one time rather than repeated. In the paper, a
convolution model of forgetting curve is proposed to simulate the memory
process during learning. In this model, the memory ability (i.e. the central
procedure in the working memory model) and learning material (i.e. the input in
the working memory model) is regarded as the system function and the input
function, respectively. The status of forgetting (i.e. the output in the
working memory model) is regarded as output function or the convolution result
of the memory ability and learning material. The model is applied to simulate
the forgetting curves in different situations. The results show that the model
is able to simulate the forgetting curves not only in one time learning
condition but also in multi-times condition. The model is further verified in
the experiments of Mandarin tone learning for Japanese learners. And the
predicted curve fits well on the test points.
| 1 | 0 | 0 | 0 | 1 | 0 |
Efficient computation of multidimensional theta functions | An important step in the efficient computation of multi-dimensional theta
functions is the construction of appropriate symplectic transformations for a
given Riemann matrix assuring a rapid convergence of the theta series. An
algorithm is presented to approximately map the Riemann matrix to the Siegel
fundamental domain. The shortest vector of the lattice generated by the Riemann
matrix is identified exactly, and the algorithm ensures that its length is
larger than $\sqrt{3}/2$. The approach is based on a previous algorithm by
Deconinck et al. using the LLL algorithm for lattice reductions. Here, the LLL
algorithm is replaced by exact Minkowski reductions for small genus and an
exact identification of the shortest lattice vector for larger values of the
genus.
| 0 | 1 | 1 | 0 | 0 | 0 |
The careless use of language in quantum information | An imperative aspect of modern science is that scientific institutions act
for the benefit of a common scientific enterprise, rather than for the personal
gain of individuals within them. This implies that science should not
perpetuate existing or historical unequal social orders. Some scientific
terminology, though, gives a very different impression. I will give two
examples of terminology invented recently for the field of quantum information
which use language associated with subordination, slavery, and racial
segregation: 'ancilla qubit' and 'quantum supremacy'.
| 0 | 1 | 0 | 0 | 0 | 0 |
Origin of meteoritic stardust unveiled by a revised proton-capture rate of $^{17}$O | Stardust grains recovered from meteorites provide high-precision snapshots of
the isotopic composition of the stellar environment in which they formed.
Attributing their origin to specific types of stars, however, often proves
difficult. Intermediate-mass stars of 4-8 solar masses are expected to
contribute a large fraction of meteoritic stardust. However, no grains have
been found with characteristic isotopic compositions expected from such stars.
This is a long-standing puzzle, which points to serious gaps in our
understanding of the lifecycle of stars and dust in our Galaxy. Here we show
that the increased proton-capture rate of $^{17}$O reported by a recent
underground experiment leads to $^{17}$O/$^{16}$O isotopic ratios that match
those observed in a population of stardust grains, for proton-burning
temperatures of 60-80 million K. These temperatures are indeed achieved at the
base of the convective envelope during the late evolution of intermediate-mass
stars of 4-8 solar masses, which reveals them as the most likely site of origin
of the grains. This result provides the first direct evidence that these stars
contributed to the dust inventory from which the Solar System formed.
| 0 | 1 | 0 | 0 | 0 | 0 |
Some Connections Between Cycles and Permutations that Fix a Set and Touchard Polynomials and Covers of Multisets | We present a new proof of a fundamental result concerning cycles of random
permutations which gives some intuition for the connection between Touchard
polynomials and the Poisson distribution. We also introduce a rather novel
permutation statistic and study its distribution. This quantity, indexed by
$m$, is the number of sets of size $m$ fixed by the permutation. This leads to
a new and simpler derivation of the exponential generating function for the
number of covers of certain multisets.
| 0 | 0 | 1 | 0 | 0 | 0 |
Generalizing the first-difference correlated random walk for marine animal movement data | Animal telemetry data are often analysed with discrete time movement models
assuming rotation in the movement. These models are defined with equidistant
distant time steps. However, telemetry data from marine animals are observed
irregularly. To account for irregular data, a time-irregularised
first-difference correlated random walk model with drift is introduced. The
model generalizes the commonly used first-difference correlated random walk
with regular time steps by allowing irregular time steps, including a drift
term, and by allowing different autocorrelation in the two coordinates. The
model is applied to data from a ringed seal collected through the Argos
satellite system, and is compared to related movement models through
simulations. Accounting for irregular data in the movement model results in
accurate parameter estimates and reconstruction of movement paths. Measured by
distance, the introduced model can provide more accurate movement paths than
the regular time counterpart. Extracting accurate movement paths from uncertain
telemetry data is important for evaluating space use patterns for marine
animals, which in turn is crucial for management. Further, handling irregular
data directly in the movement model allows efficient simultaneous analysis of
several animals.
| 0 | 0 | 0 | 0 | 1 | 0 |
Low spin wave damping in the insulating chiral magnet Cu$_{2}$OSeO$_{3}$ | Chiral magnets with topologically nontrivial spin order such as Skyrmions
have generated enormous interest in both fundamental and applied sciences. We
report broadband microwave spectroscopy performed on the insulating chiral
ferrimagnet Cu$_{2}$OSeO$_{3}$. For the damping of magnetization dynamics we
find a remarkably small Gilbert damping parameter of about $1\times10^{-4}$ at
5 K. This value is only a factor of 4 larger than the one reported for the best
insulating ferrimagnet yttrium iron garnet. We detect a series of sharp
resonances and attribute them to confined spin waves in the mm-sized samples.
Considering the small damping, insulating chiral magnets turn out to be
promising candidates when exploring non-collinear spin structures for high
frequency applications.
| 0 | 1 | 0 | 0 | 0 | 0 |
Adversarial PoseNet: A Structure-aware Convolutional Network for Human Pose Estimation | For human pose estimation in monocular images, joint occlusions and
overlapping upon human bodies often result in deviated pose predictions. Under
these circumstances, biologically implausible pose predictions may be produced.
In contrast, human vision is able to predict poses by exploiting geometric
constraints of joint inter-connectivity. To address the problem by
incorporating priors about the structure of human bodies, we propose a novel
structure-aware convolutional network to implicitly take such priors into
account during training of the deep network. Explicit learning of such
constraints is typically challenging. Instead, we design discriminators to
distinguish the real poses from the fake ones (such as biologically implausible
ones). If the pose generator (G) generates results that the discriminator fails
to distinguish from real ones, the network successfully learns the priors.
| 1 | 0 | 0 | 0 | 0 | 0 |
Deep Learning from Shallow Dives: Sonar Image Generation and Training for Underwater Object Detection | Among underwater perceptual sensors, imaging sonar has been highlighted for
its perceptual robustness underwater. The major challenge of imaging sonar,
however, arises from the difficulty in defining visual features despite limited
resolution and high noise levels. Recent developments in deep learning provide
a powerful solution for computer-vision researches using optical images.
Unfortunately, deep learning-based approaches are not well established for
imaging sonars, mainly due to the scant data in the training phase. Unlike the
abundant publically available terrestrial images, obtaining underwater images
is often costly, and securing enough underwater images for training is not
straightforward. To tackle this issue, this paper presents a solution to this
field's lack of data by introducing a novel end-to-end image-synthesizing
method in the training image preparation phase. The proposed method present
image synthesizing scheme to the images captured by an underwater simulator.
Our synthetic images are based on the sonar imaging models and noisy
characteristics to represent the real data obtained from the sea. We validate
the proposed scheme by training using a simulator and by testing the simulated
images with real underwater sonar images obtained from a water tank and the
sea.
| 1 | 0 | 0 | 0 | 0 | 0 |
Large-scale Feature Selection of Risk Genetic Factors for Alzheimer's Disease via Distributed Group Lasso Regression | Genome-wide association studies (GWAS) have achieved great success in the
genetic study of Alzheimer's disease (AD). Collaborative imaging genetics
studies across different research institutions show the effectiveness of
detecting genetic risk factors. However, the high dimensionality of GWAS data
poses significant challenges in detecting risk SNPs for AD. Selecting relevant
features is crucial in predicting the response variable. In this study, we
propose a novel Distributed Feature Selection Framework (DFSF) to conduct the
large-scale imaging genetics studies across multiple institutions. To speed up
the learning process, we propose a family of distributed group Lasso screening
rules to identify irrelevant features and remove them from the optimization.
Then we select the relevant group features by performing the group Lasso
feature selection process in a sequence of parameters. Finally, we employ the
stability selection to rank the top risk SNPs that might help detect the early
stage of AD. To the best of our knowledge, this is the first distributed
feature selection model integrated with group Lasso feature selection as well
as detecting the risk genetic factors across multiple research institutions
system. Empirical studies are conducted on 809 subjects with 5.9 million SNPs
which are distributed across several individual institutions, demonstrating the
efficiency and effectiveness of the proposed method.
| 1 | 0 | 0 | 1 | 0 | 0 |
Strong Functional Representation Lemma and Applications to Coding Theorems | This paper shows that for any random variables $X$ and $Y$, it is possible to
represent $Y$ as a function of $(X,Z)$ such that $Z$ is independent of $X$ and
$I(X;Z|Y)\le\log(I(X;Y)+1)+4$ bits. We use this strong functional
representation lemma (SFRL) to establish a bound on the rate needed for
one-shot exact channel simulation for general (discrete or continuous) random
variables, strengthening the results by Harsha et al. and Braverman and Garg,
and to establish new and simple achievability results for one-shot
variable-length lossy source coding, multiple description coding and Gray-Wyner
system. We also show that the SFRL can be used to reduce the channel with state
noncausally known at the encoder to a point-to-point channel, which provides a
simple achievability proof of the Gelfand-Pinsker theorem.
| 1 | 0 | 1 | 0 | 0 | 0 |
Emergence of Structured Behaviors from Curiosity-Based Intrinsic Motivation | Infants are experts at playing, with an amazing ability to generate novel
structured behaviors in unstructured environments that lack clear extrinsic
reward signals. We seek to replicate some of these abilities with a neural
network that implements curiosity-driven intrinsic motivation. Using a simple
but ecologically naturalistic simulated environment in which the agent can move
and interact with objects it sees, the agent learns a world model predicting
the dynamic consequences of its actions. Simultaneously, the agent learns to
take actions that adversarially challenge the developing world model, pushing
the agent to explore novel and informative interactions with its environment.
We demonstrate that this policy leads to the self-supervised emergence of a
spectrum of complex behaviors, including ego motion prediction, object
attention, and object gathering. Moreover, the world model that the agent
learns supports improved performance on object dynamics prediction and
localization tasks. Our results are a proof-of-principle that computational
models of intrinsic motivation might account for key features of developmental
visuomotor learning in infants.
| 0 | 0 | 0 | 1 | 0 | 0 |
Predicting shim gaps in aircraft assembly with machine learning and sparse sensing | A modern aircraft may require on the order of thousands of custom shims to
fill gaps between structural components in the airframe that arise due to
manufacturing tolerances adding up across large structures. These shims are
necessary to eliminate gaps, maintain structural performance, and minimize
pull-down forces required to bring the aircraft into engineering nominal
configuration for peak aerodynamic efficiency. Gap filling is a time-consuming
process, involving either expensive by-hand inspection or computations on vast
quantities of measurement data from increasingly sophisticated metrology
equipment. Either case amounts to significant delays in production, with much
of the time spent in the critical path of aircraft assembly. This work presents
an alternative strategy for predictive shimming, based on machine learning and
sparse sensing to first learn gap distributions from historical data, and then
design optimized sparse sensing strategies to streamline data collection and
processing. This new approach is based on the assumption that patterns exist in
shim distributions across aircraft, which may be mined and used to reduce the
burden of data collection and processing in future aircraft. Specifically,
robust principal component analysis is used to extract low-dimensional patterns
in the gap measurements while rejecting outliers. Next, optimized sparse
sensors are obtained that are most informative about the dimensions of a new
aircraft in these low-dimensional principal components. We demonstrate the
success of the proposed approach, called PIXel Identification Despite
Uncertainty in Sensor Technology (PIXI-DUST), on historical production data
from 54 representative Boeing commercial aircraft. Our algorithm successfully
predicts $99\%$ of shim gaps within the desired measurement tolerance using
$3\%$ of the laser scan points typically required; all results are
cross-validated.
| 0 | 0 | 0 | 1 | 0 | 0 |
Multivariate central limit theorems for Rademacher functionals with applications | Quantitative multivariate central limit theorems for general functionals of
possibly non-symmetric and non-homogeneous infinite Rademacher sequences are
proved by combining discrete Malliavin calculus with the smart path method for
normal approximation. In particular, a discrete multivariate second-order
Poincaré inequality is developed. As a first application, the normal
approximation of vectors of subgraph counting statistics in the
Erdős-Rényi random graph is considered. In this context, we further
specialize to the normal approximation of vectors of vertex degrees. In a
second application we prove a quantitative multivariate central limit theorem
for vectors of intrinsic volumes induced by random cubical complexes.
| 0 | 0 | 1 | 0 | 0 | 0 |
Modification of Social Dominance in Social Networks by Selective Adjustment of Interpersonal Weights | According to the DeGroot-Friedkin model of a social network, an individual's
social power evolves as the network discusses individual opinions over a
sequence of issues. Under mild assumptions on the connectivity of the network,
the social power of every individual converges to a constant strictly positive
value as the number of issues discussed increases. If the network has a special
topology, termed "star topology", then all social power accumulates with the
individual at the centre of the star. This paper studies the strategic
introduction of new individuals and/or interpersonal relationships into a
social network with star topology to reduce the social power of the centre
individual. In fact, several strategies are proposed. For each strategy, we
derive necessary and sufficient conditions on the strength of the new
interpersonal relationships, based on local information, which ensures that the
centre individual no longer has the greatest social power within the social
network. Interpretations of these conditions show that the strategies are
remarkably intuitive and that certain strategies are favourable compared to
others, all of which is sociologically expected.
| 1 | 0 | 0 | 0 | 0 | 0 |
On Abrikosov Lattice Solutions of the Ginzburg-Landau Equations | We prove existence of Abrikosov vortex lattice solutions of the
Ginzburg-Landau equations of superconductivity, with multiple magnetic flux
quanta per a fundamental cell. We also revisit the existence proof for the
Abrikosov vortex lattices, streamlining some arguments and providing some
essential details missing in earlier proofs for a single magnetic flux quantum
per a fundamental cell.
| 0 | 0 | 1 | 0 | 0 | 0 |
$0.7-2.5~μ$m spectra of Hilda asteroids | The Hilda asteroids are primitive bodies in resonance with Jupiter whose
origin and physical properties are not well understood. Current models posit
that these asteroids formed in the outer Solar System and were scattered along
with the Jupiter Trojans into their present-day positions during a chaotic
episode of dynamical restructuring. In order to explore the surface composition
of these enigmatic objects in comparison with an analogous study of Trojans
(Emery et al. 2011), we present new near-infrared spectra (0.7-2.5 $\mu$m) of
25 Hilda asteroids. No discernible absorption features are apparent in the
data. Synthesizing the bimodalities in optical color and infrared reflectivity
reported in previous studies, we classify 26 of the 28 Hildas in our spectral
sample into the so-called less-red and red sub-populations and find that the
two sub-populations have distinct average spectral shapes. Combining our
results with visible spectra, we find that Trojans and Hildas possess similar
overall spectral shapes, suggesting that the two minor body populations share a
common progenitor population. A more detailed examination reveals that while
the red Trojans and Hildas have nearly identical spectra, less-red Hildas are
systematically bluer in the visible and redder in the near-infrared than
less-red Trojans, indicating a putative broad, shallow absorption feature
between 0.5 and 1.0 $\mu$m. We argue that the less-red and red objects found in
both Hildas and Trojans represent two distinct surface chemistries and
attribute the small discrepancy between less-red Hildas and Trojans to the
difference in surface temperatures between the two regions.
| 0 | 1 | 0 | 0 | 0 | 0 |
Ground-state properties of unitary bosons: from clusters to matter | The properties of cold Bose gases at unitarity have been extensively
investigated in the last few years both theoretically and experimentally. In
this paper we use a family of interactions tuned to two-body unitarity and very
weak three-body binding to demonstrate the universal properties of both
clusters and matter. We determine the universal properties of finite clusters
up to 60 particles and, for the first time, explicitly demonstrate the
saturation of energy and density with particle number and compare with bulk
properties. At saturation in the bulk we determine the energy, density, two-
and three-body contacts and the condensate fraction. We find that uniform
matter is more bound than three-body clusters by nearly two orders of
magnitude, the two-body contact is very large in absolute terms, and yet the
condensate fraction is also very large, greater than 90%. Equilibrium
properties of these systems may be experimentally accessible through rapid
quenching of weakly-interacting boson superfluids.
| 0 | 1 | 0 | 0 | 0 | 0 |
Lower bounds on the Noether number | The best known method to give a lower bound for the Noether number of a given
finite group is to use the fact that it is greater than or equal to the Noether
number of any of the subgroups or factor groups. The results of the present
paper show in particular that these inequalities are strict for proper
subgroups or factor groups. This is established by studying the algebra of
coinvariants of a representation induced from a representation of a subgroup.
| 0 | 0 | 1 | 0 | 0 | 0 |
A Survey on the Adoption of Cloud Computing in Education Sector | Education is a key factor in ensuring economic growth, especially for
countries with growing economies. Today, students have become more
technologically savvy as teaching and learning uses more advance technology day
in, day out. Due to virtualize resources through the Internet, as well as
dynamic scalability, cloud computing has continued to be adopted by more
organizations. Despite the looming financial crisis, there has been increasing
pressure for educational institutions to deliver better services using minimal
resources. Leaning institutions, both public and private can utilize the
potential advantage of cloud computing to ensure high quality service
regardless of the minimal resources available. Cloud computing is taking a
center stage in academia because of its various benefits. Various learning
institutions use different cloud-based applications provided by the service
providers to ensure that their students and other users can perform both
academic as well as business-related tasks. Thus, this research will seek to
establish the benefits associated with the use of cloud computing in learning
institutions. The solutions provided by the cloud technology ensure that the
research and development, as well as the teaching is more sustainable and
efficient, thus positively influencing the quality of learning and teaching
within educational institutions. This has led to various learning institutions
adopting cloud technology as a solution to various technological challenges
they face on a daily routine.
| 1 | 0 | 0 | 0 | 0 | 0 |
Deep Neural Networks as Gaussian Processes | It has long been known that a single-layer fully-connected neural network
with an i.i.d. prior over its parameters is equivalent to a Gaussian process
(GP), in the limit of infinite network width. This correspondence enables exact
Bayesian inference for infinite width neural networks on regression tasks by
means of evaluating the corresponding GP. Recently, kernel functions which
mimic multi-layer random neural networks have been developed, but only outside
of a Bayesian framework. As such, previous work has not identified that these
kernels can be used as covariance functions for GPs and allow fully Bayesian
prediction with a deep neural network.
In this work, we derive the exact equivalence between infinitely wide deep
networks and GPs. We further develop a computationally efficient pipeline to
compute the covariance function for these GPs. We then use the resulting GPs to
perform Bayesian inference for wide deep neural networks on MNIST and CIFAR-10.
We observe that trained neural network accuracy approaches that of the
corresponding GP with increasing layer width, and that the GP uncertainty is
strongly correlated with trained network prediction error. We further find that
test performance increases as finite-width trained networks are made wider and
more similar to a GP, and thus that GP predictions typically outperform those
of finite-width networks. Finally we connect the performance of these GPs to
the recent theory of signal propagation in random neural networks.
| 1 | 0 | 0 | 1 | 0 | 0 |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.