sentence1
stringlengths 16
446
| sentence2
stringlengths 14
436
|
|---|---|
extensive attention-based deep models have been developed for images . | so far , attention-based deep architectures has been proposed for image . |
recently , with the availability of powerful gpus and large amounts of labeled training data , deep convolutional neural networks have demonstrated impressive performances for many computer vision tasks such as object recognition . | in recent years , many models based on convolutional neural networks have been proposed and used for various computer vision tasks such as image classification . |
many improvements for dcf tracking approaches have also been proposed , such as samf to mitigate boundary effects . | many improvements for dcf tracking approaches have been proposed , such as samf to mitigate boundary effects . |
convolutional neural networks have achieved impressive state-of-the-art results on image classification . | convolutional neural networks algorithms have been achieving outstanding classification capabilities in several complex tasks such as image recognition . |
in the recent years , non-gaussian states have been recognized as a valuable resource for several quantum information protocols . | these nongaussian quantum operations have proven advantageous in many scenarios such as entanglement enhancement . |
reconstruction techniques for classical inverse sturm-liouville problems . | inverse spectral problem for the sturm-liouville equation . |
complex structures with dense γ-orbits are called ergodic . | ergodic and there is a temperature that can be associated with them . |
for instance , valgaerts et al leveraged the fundamental matrix prior as an additional weak prior within a variational framework . | wedel et al utilized fundamental matrix prior as a weak constraint in a variational framework . |
this modification consists of the following local change of the exchange constants . | the simplest modification is a noncovariant frequency cutoff . |
deep neural networks have gained popularity in recent years thanks to their achievements in many applications including computer vision , signal and image processing , speech recognition . | in recent years , deep neural networks have gained enormous popularity for a wide spectrum of applications , ranging from image recognition . |
hasan et al in used an auto-encoder to learn appearance based representation under reconstruction framework . | hasan et al proposed a fully convolutional feedforward deep auto-encoder based approach to learn local features and classifier . |
dropout is one of the regularization techniques that is used widely to reduce overfitting in deep neural networks . | dropout is a technique that reduces overfitting by randomly dropping out neurons during training , thus forcing the network to learn redundantly . |
recently , deep reinforcement learning approaches have gained considerable attention due to their success in learning efficient policies to play games . | recent research has shown that deep learning can be applied to learn a useful representation for reinforcement learning problems . |
reinforcement learning is a machine learning approach that enables intelligent agents to learn the optimal behavior via trail-and-error . | reinforcement learning is the branch of machine learning in which artificial agents have to learn an optimal behavior while interacting with a particular environment . |
for higher order qam beyond qpsk , the necessary changes for our joint sdr receiver include the relaxed box constraints for diagonal elements and the symbol-to-bit mapping constraints . | for higher order qam beyond qpsk , the necessary changes for our joint sdr receiver include box relaxation of diagonal elements of x k and the modification of symbol-to-bit mapping constraints . |
deep neural networks have shown improvement in state-of-the-art in different tasks , such as image classification . | deep neural networks have demonstrated dramatically accurate results for challenging tasks . |
in this work , a convolutional layer is by default followed by batch normalization activation unit . | batch normalization is introduced following each convolutional layer . |
convolutional neural networks have recently achieved the state-of-the-art performance in many image analysis tasks . | deep neural networks have shown remarkable success in many domains , such as computer vision . |
fattal proposed a haze removal algorithm which was based on the observation of a generic regularity in natural images in which pixels of small image patches exhibit one dimensional distribution in rgb color space . | fattal introduced a method that exploits the observation that pixels of small image patches typically exhibit a one-dimensional distribution in rgb color space , named color lines . |
details on this very successful tdlda-md approach for free clusters can be found in . | details on the successful tdlda-md approach for free clusters can be found in . |
the γ term denotes straining due to the time-dependence of the metric . | and the δ term is the price to pay for this deviation . |
also , generative adversarial networks have been used for synthetic data augmentation . | methods like variational auto-encoders and generative adversarial networks have found success at modeling data distributions . |
variational auto-encoder and generative adversarial networks show good performance in modelling real-world data such as images well . | deep generative models such as generative adversarial network and variational auto-encoder show remarkable performance in learning the manifold structure of the data . |
for the data used here , only the second case can give the real constraint on distance and the host galaxy extinction . | thus the method proposed in this section can indeed give information on the distance and its host galaxy extinction . |
the adam optimizer was used with an initial learning rate 1e-4 , which decays during training . | adam optimizer was employed as the optimizer with a initial learning rate of 1e-4 . |
the great success of deep learning network in 2d image detection has accelerated the development of 3d object detection techniques . | with the development and application of deep learning in computer vision , object detection and recognition has been studied . |
we train the hyper-parameters of the network using minibatch training using the adam update rule . | we use the adam optimizer to adjust the learning rate for both the policy and value function networks . |
in this paper , we make frequent use of a connected component identification algorithm due to thurimella . | for this representation , thurimella gave a distributed connected components algorithm in otime . |
we have studied effects of the holonomy corrections to the equation for tensor modes . | we consider the influence of the holonomy corrections to the equation for tensor modes . |
csorba showed that for every finite free z 2 -complex x , there is a graph g such that homand x are z 2 -homotopy equivalent . | braun showed that for every z 2 -cw-complex x , there is a simple graph g such that homand x are z 2 -homotopy equivalent . |
monolayer transition metal dichalcogenides have recently attracted great interest in the field of photonics because of their distinctive optical and spin properties . | monolayer transition metal dichalcogenides have recently emerged as interesting candidates for optoelectronic applications due to their unique optical properties . |
the correlation matrix therefore explicitly depends on the model parameters . | however , most of the important parameters are not linearly or polynomially dependent on the correlation matrix . |
however , gravity , which is a property of spacetime itself , must be free to propagate in the bulk . | gravity is a natural candidate to mediate supersymmetry breaking because regardless of other details of the model , the gravitational interaction between the hidden and the observed sector exists . |
the success of deep neural networks in a wide variety of computer vision tasks has emphasized the need for highly non-linear and nonparametric models . | deep convolutional neural networks have recently become increasingly important for computer vision applications . |
deep convolutional neural networks have made significant breakthroughs in many visual understanding tasks including image classification . | d eep neural networks have advanced to show the state-of-the-art performance in many of computer vision applications , such as image classification . |
mathematics is a fundamentally creative process in a finite but perhaps . | mathematics is the science of skillful operations with concepts and rules invented for just this purpose . |
understanding the stability of ecological networks is of pivotal importance in theoretical biology . | understanding the origin of and maintaining biodiversity is of obvious paramount importance in ecology and biology . |
deep neural networks have demonstrated impressive performance on many hard perception problems . | convolutional neural networks have had huge successes since krizhevsky et al , especially in computer vision applications . |
the calibrated visibility data were fourier-transformed and cleaned with miriad to produce images . | the calibrated data were then imaged and analyzed in the standard manner using the miriad and karma softwares . |
more precisely , we will be interested in the lie algebras sln and their universal central extensions . | we will be interested in the structure of the lie algebras sln and of their universal central extensions . |
reinforcement learning is a powerful framework which allows an agent to behave near-optimally through a trial and error exploration of the environment . | reinforcement learning aims at automatically learning an optimal strategy from the sequential interactions between the agent and the environment by trial-and-error . |
in proceedings of the thirteenth national conference on arti cial intel li gence , pp . | in proceedings of the eighth national conference on arti cial intel ligence , pp . |
socher et al use neural networks for building sentence and image vector representations that are then mapped into a common embedding space . | socher et al embed a fullframe cnn representation with the sentence representation from a semantic dependency tree recursive neural network . |
convolutional neural network has achieved great success in image recognition and object detection . | recently deep neural networks have attained impressive performance in many fields such as image classification . |
as an alternative to compression , hinton et al proposed using soft-targets to transfer knowledge from a costly ensemble to a single model while largely preserving prediction accuracy . | hinton et al proposed the knowledge distillation approach to compress the knowledge of a large and computational expensive model to a single computational efficient neural network . |
multi-task learning has been successfully used to improve performance in various tasks , including machine translation and image captioning . | using multiple encoders has been shown to be useful in mutli-task learning , multi-source translation and reading comprehension . |
the root node is the single node which has no father and is located at the top of the tree , while the leaf nodes are all the nodes which have no children . | the root node is a random vector r , internal nodes stand for vectors formed by application of a matrix on a vector , and leaves . |
numerical simulations have indicated that vortices could also account for phenomena related to chiral symmetry , such as causing topological charge fluctuations and spontaneous chiral symmetry breaking . | in addition , simulations have indicated that vortices could also account for phenomena related to chiral symmetry , such as topological charge and spontaneous chiral symmetry breaking . |
in the shuttling regime , electron transport is highly deterministic characterized by the extraordinary sub-poissonian fano factor . | in the semiclassical case , electron transport through a bistable coexistent channels of shutting and tunneling causes super-poissonian noise spectrum both at zero . |
variational autoencoders are generative models that can map from random noise to meaningful manifolds . | variational autoencoders are the generative models that are capable of learning approximated data distribution by applying variational inference . |
generative adversarial networks are one of the main approaches to learning such models in a fully unsupervised fashion . | such networks can be trained in an adversarial manner , as described in generative adversarial networks . |
in the authors use a fusion of smri and pet images together with canonical correlation analysis . | in authors use a fusion of smri and pet images together with canonical correlation analysis . |
the world-sheet description of the non-critical string theories is given by the twodimensional quantum gravity . | two dimensional non-critical string theory has a non-perturbative description in terms of the singlet sector of the quantum mechanics of a single hermitian matrix . |
first proposed by in 2014 , generative adversarial networks have been quite successful at generating realistic images . | in particular , generative adversarial networks achieve very impressive results in synthesizing individual images . |
a clique is a set of vertices for which no node outside the clique is connected to all members of the clique . | a homogeneous set for this coloring is called a clique in x . |
the upper plot shows the corresponding peak amplitude of the resonant spectrum . | the lower plot shows the modulation of the resonant frequency with external magnetic field . |
asterisks denote non-oscillatory , diamonds oscillatory solutions . | asterisks denote fr i sources , diamonds are fr ii sources , and the solid line shows the best fit to the fr i luminosity correlation . |
topological defects are believed to be relevant for structure formation in the universe . | these objects have been investigated in the context of cosmology to explain the formation of large structure in the universe . |
for this study we ignore training hyper-parameters and use the adam optimizer . | for model training , we use the adam optimizer with an initial learning rate of 10 -4 . |
cheng et al modeled job hop activities to rank influential companies . | cheng et al modeled job transition activities to rank influential companies . |
neural networks have been widely used in many natural language processing tasks , including neural machine translation , text summarization and dialogue systems . | recurrent neural networks are widely used for sequence modelling tasks in domains such as natural language processing , speech recognition , and reinforcement learning . |
to reduce the parameter size , denten et al applied the lowrank approximation approach to compress the neural networks with linear structures . | denton et al use low rank approximation and clustering techniques to approximate a single convolutional layer . |
we can then tap into our recent work on modified-cs which solves the sparse recovery problem when partial support knowledge is available . | we then tap into our recent work on modified-cs which solves the sparse recovery problem with much fewer measurements when reliable support knowledge is available . |
despite the great success , deep neural networks have been shown to be highly vulnerable to adversarial examples . | however , recent research has shown that well-trained deep neural networks are rather vulnerable to adversarial examples . |
this duality can be explained as a geometric flop in m theory on a g 2 holonomy manifold . | these features arise in the effective theory if we compactify on a singular g 2 holonomy manifold . |
garg and tamassia showed that it is n p-complete to decide whether a 4-planar graph admits an orthogonal drawing without any edge bends . | garg and tamassia show that it is n p-hard to decide whether a 4-planar graph admits an orthogonal drawing without any bends . |
in practice , however , the best known algorithm still has an o-time worstcase bound but uses several clever tricks when compared to the brute-force algorithm . | in practice , however , the best known algorithm still has an o-time worst-case bound but uses several clever tricks when compared to the straightforward brute-force algorithm . |
complex models such as deep neural networks have shown remarkable success in applications such as computer vision , speech and time series analysis . | deep neural networks have shown improvement in state-of-the-art in different tasks , such as image classification . |
goodfellow et al proposed generative adversarial nets for estimating generative models with an adversarial process . | goodfellow et al proposed generative adversarial networks that greatly improve image generation quality . |
diameter is the average number of hops between every pair of vertices . | the diameter is the mean of measured diameters , if available , or the mean of estimated diameters . |
recent progress in string theory has stimulated interest in solitons in noncommutative field theories . | recently there has been considerable interest in solitons in noncommutative field theories , the main motivation coming from string theory . |
for completeness we give the analytic expression for these corrections in appendix a . | we therefore omit this contribution in what follows but for completeness give the analytic expression for it in appendix a . |
in order to determine the interlayer distance we carry out dft calculations with the generalized gradient approximation in the parametrization of perdew , burke and ernzerhof which combines the nonequilibrium green function method and ncdft . | the electron-core interactions are described by the projector augmented wave method , and we use perdew-burke-ernzerhof parametrization of the generalized gradient approximation for the exchange-correlation functional . |
topological structures appear in a diversity of contexts in high energy physics . | topological structures appear in high energy physics in a variety of dimensions . |
since the laser is a poisson light source , we expect this to be the case . | laser is a measure of the atom number fluctuations only . |
the generative adversarial network is a powerful generative model that can generate plausible images . | this gives rise to a family of models known as generative adversarial networks . |
we use batch normalisation layers after every convolutional layer . | we use a simple fully-connected layer with batch normalization to project the input . |
if the gravitino is the lightest supersymmetric particle , it is stable and therefore represents a good dark matter candidate . | when the gravitino is the lsp , there are number of new implications of supersymmetry for cosmology . |
for more information on the blow-up criteria of compressible navier-stokes flows , we refer to and the references therein . | for more information on the blowup criteria of barotropic compressible flow , we refer to and the references therein . |
deep neural networks have recently been achieved breakthroughs in several domains such as computer vision . | convolutional neural networks have recently been applied to various computer vision tasks such as image classification . |
this confirms that it is a good measure of frameness . | this well confirms that algorithm 1 is a fast method . |
during photoemission , the electrons are non-boltzmann distributed in the vicinity of the grain . | in this case , ions are slowed down and deflected in front of the grain . |
devise trains a linear mapping between visual and semantic fs by an effective ranking loss formulation . | devise employs a ranking formulation for zero-shot learning using images and distributed text representations . |
convolutional neural networks have greatly advanced the state of the art in all those structured output tasks . | deep neural networks have achieved state-of-the-art performance on a wide variety of machine learning tasks . |
in parentheses is the significance of a non-zero slope followed by the number of sn included in the fit . | in parentheses is the significance followed by the number of sn included in the fit . |
the ellipses denote half-integral spin higher states . | the ellipses denote terms involving other representations and also terms involving three fields . |
deep learning has achieved widespread success in a variety of settings , such as computer vision . | deep learning using convolutional neural networks has achieved excellent performance for a wide range of tasks , such as image recognition . |
sandage , ar , in galaxies and the universe , eds . | heckman , tm , in paired and interacting galaxies , eds . |
the youtube faces dataset exemplify both unconstained and controlled video settings . | the youtube faces dataset exemplify both unconstrained and controlled video settings . |
first , we show that states of distinguishable particles typically are not useful for metrology , despite having a large amount of entanglement as measured by the entanglement entropy . | first , we show that haar-random isospectral states of distinguishable particles are typically not useful for quantum metrology even if they are pure and hence typically highly entangled . |
the composite multiple reciprocity approach is also very useful in the kernel . | recursive multiple reciprocity method can become very efficient via the scale . |
quantum mechanics is a one dimensional field theory . | if quantum mechanics is a complete description then this eminently plausible postulate must be incorrect . |
the genus of the handlebody is the genus of its boundary . | the result of adding g of them is called a genus g handlebody . |
inflationary cosmology , the epoch of quasi-de sitter expansion in the very early universe , relieves the standard big bang cosmology from the requirement of extremely finely tuned initial conditions . | inflationary cosmology is the leading paradigm of the very early universe , in which the universe has experienced a primordial phase of quasi-de sitter expansion . |
the expectation-maximization algorithm can be used to obtain maximum likelihood estimates . | a common approach is the method of maximum likelihood for which the expectation-maximization algorithm can be applied . |
convolutional neural networks , cnn , have recently achieved state of the art performance in a number of computer vision tasks . | deep convolutional neural networks have led to major breakthroughs in many computer vision tasks . |
recently , chimera states have been studied in nonlocally coupled type-i excitable systems . | previously , chimera states have been found in one-layer networks consisting of coupled oscillatory 46 and excitatory fhn systems . |
this line of work was pioneered by aaron wyner , who introduced the wiretap channel and established fundamental results of creating perfectly secure communications without relying on private keys . | the idea was pioneered by wyner , who introduced the wiretap channel and established the possibility of creating perfectly secure communication links without relying on private keys . |
in the current era , deep learning techniques have made significant development in fields such as artificial intelligence . | over the past few years , deep neural networks have driven advances in many practical problems , such as image classification . |
an automorphism of a bipartite graph x with a given black-and-white coloring that maps black vertices to white vertices is called a duality . | the duality is a trivial geometrical reflection . |
Subsets and Splits