text
stringlengths
27
153
label
stringlengths
27
153
id
int64
0
40k
Accelerating network such received increased has neural attention. over-parameterized
Accelerating such over-parameterized neural network has received increased attention.
38,700
a three-stage A retraining. algorithm pruning pipeline, and typical i.e., pruning, training, is
A typical pruning algorithm is a three-stage pipeline, i.e., training, pruning, and retraining.
38,701
update filters stage. during pruned the the retraining First, we
First, we update the pruned filters during the retraining stage.
38,702
has model this data. learn to way, In training the capacity the from enough
In this way, the model has enough capacity to learn from the training data.
38,703
during at We filters training few and first the prune more filters prune procedure. asymptotically
We prune few filters at first and asymptotically prune more filters during the training procedure.
38,704
ASFP classification the Experiments show on benchmarks. image our of effectiveness
Experiments show the effectiveness of our ASFP on image classification benchmarks.
38,705
are latency challenges reducing LTE the when Design shown. faced in
Design challenges faced when reducing the latency in LTE are shown.
38,706
short structure time of analyzed. and The frame enhancements are performance processing
The performance of short processing time and frame structure enhancements are analyzed.
38,707
LTE DCI evaluated. increase Our to and channel Duplication proposed is reliability method control presented
Our proposed DCI Duplication method to increase LTE control channel reliability is presented and evaluated.
38,708
models. statistical based on The is three proposed method
The proposed method is based on three statistical models.
38,709
All scans data computed tomography automatically from head scans. (CT) and face derived optical are
All data are automatically derived from computed tomography (CT) head scans and optical face scans.
38,710
fit spheres. of the head model all Finally, union to we a
Finally, we fit a head model to the union of all spheres.
38,711
our face CT head process scan. an anonymized recovery using validate We
We validate our face recovery process using an anonymized head CT scan.
38,712
a the time cloudy of or day (e.g. in day)
in a cloudy day) or time of the day (e.g.
38,713
Practice Under of consideration Programming Theory (TPLP). and in Logic
Under consideration in Theory and Practice of Logic Programming (TPLP).
38,714
leaps? always or can evolution make it gradual Is
Is evolution always gradual or can it make leaps?
38,715
that an adaptive mutagenesis hypothesize evolvable is strategy. We stress-induced
We hypothesize that stress-induced mutagenesis is an evolvable adaptive strategy.
38,716
asymptotic determine behavior and carrier-phase exact both the We positioning. for pseudo-range
We determine the exact asymptotic behavior for both pseudo-range and carrier-phase positioning.
38,717
a semantic work, we present manipulation. In this hierarchical image novel for framework
In this work, we present a novel hierarchical framework for semantic image manipulation.
38,718
generator fills the in the layout. by our textures guided Then semantic image pixel-level
Then our image generator fills in the pixel-level textures guided by the semantic layout.
38,719
new component a Hilbert introduced is Real-time A transformer as microwave Processing for (RAP). Analog
A microwave Hilbert transformer is introduced as a new component for Real-time Analog Processing (RAP).
38,720
an inference basically an not problem optimization This problem. is
This is basically an inference problem not an optimization problem.
38,721
bucket time DES supply chains. the a of and accelerate present approximate method We to
We present a time bucket method to approximate and accelerate the DES of supply chains.
38,722
by mutation Synonymous is and choice mainly selection. affected Background codon
Background Synonymous codon choice is mainly affected by mutation and selection.
38,723
demonstrate framework results achieve and performance. superior Experimental can methods with comparisons our state-of-the-art that
Experimental results and comparisons with state-of-the-art methods demonstrate that our framework can achieve superior performance.
38,724
benefits There and potential these are applications many ideas. of
There are many potential benefits and applications of these ideas.
38,725
unnoticeable often implies failures and such unavoidable of in presence (OOD) input. However, assumption out-of-distribution
However, such assumption implies unavoidable and often unnoticeable failures in presence of out-of-distribution (OOD) input.
38,726
detection discriminative this issue propose input to OOD by of We address pixels data. in
We propose to address this issue by discriminative detection of OOD pixels in input data.
38,727
prediction setup. dense resolution images experiments our perform on a natural high We in
We perform our experiments on high resolution natural images in a dense prediction setup.
38,728
coprime constructed arrays quadratic from The lattices proposed ideal CRT embedded integers. are by
The proposed CRT arrays are constructed by ideal lattices embedded from coprime quadratic integers.
38,729
are provided of the Examples verify to feasibility methods. the proposed
Examples are provided to verify the feasibility of the proposed methods.
38,730
Recognition sixth challenge. in This details paper (EmotiW) Wild Emotion the the
This paper details the sixth Emotion Recognition in the Wild (EmotiW) challenge.
38,731
detail. and discussed baselines The are databases, in protocols
The databases, protocols and baselines are discussed in detail.
38,732
are language Wikipedia syllables Statistical on French of and models n-gram trained English corpora.
Statistical n-gram language models of syllables are trained on French and English Wikipedia corpora.
38,733
also and We lexicon n-gram with models. syllable models compare character the
We also compare the syllable models with lexicon and character n-gram models.
38,734
demonstrates images for task. of carefully-rendered synthetic intrinsic suprising work data Our the effectiveness the
Our work demonstrates the suprising effectiveness of carefully-rendered synthetic data for the intrinsic images task.
38,735
the efficacy our demonstrate method. Extensive of experiments
Extensive experiments demonstrate the efficacy of our method.
38,736
passing iteratively framework, Within a and computes labels. GPNN message inference node matrices the adjacency
Within a message passing inference framework, GPNN iteratively computes the adjacency matrices and node labels.
38,737
used Lens been aberrations image. provenance determine of to an previously the have
Lens aberrations have previously been used to determine the provenance of an image.
38,738
via only Markov considers saliency chain nodes. detection boundaries Traditional
Traditional saliency detection via Markov chain only considers boundaries nodes.
38,739
Markov boundaries The bidirectional chain. both proposed approach combines prior through foreground and cues
The proposed approach combines both boundaries and foreground prior cues through bidirectional Markov chain.
38,740
SE we an that blocks of Hence, can F-CNN part integral be architectures. believe future
Hence, we believe that SE blocks can be an integral part of future F-CNN architectures.
38,741
Neural Networks Convolutional (SISR) have super-resolution problem. great (CNNs) demonstrated single-image results for the
Convolutional Neural Networks (CNNs) have demonstrated great results for the single-image super-resolution (SISR) problem.
38,742
CNN and solve to algorithms models computationally promote deep Currently, expensive SISR. most
Currently, most CNN algorithms promote deep and computationally expensive models to solve SISR.
38,743
novel we relatively propose SISR However, a of computations. number less method that uses
However, we propose a novel SISR method that uses relatively less number of computations.
38,744
connections unused get that removed. On have we training, group convolutions
On training, we get group convolutions that have unused connections removed.
38,745
computations testing steps these reduce time. the significantly of number at required All
All these steps significantly reduce the number of computations required at testing time.
38,746
is input for the learning. output easier this, network with Along bicubic upsampled added to
Along with this, bicubic upsampled input is added to the network output for easier learning.
38,747
step foreground video Extracting accurate from essential for an objects is a applications. scene many
Extracting accurate foreground objects from a scene is an essential step for many video applications.
38,748
is novel innovation our key method of information The technique. fusion a
The key innovation of our method is a novel information fusion technique.
38,749
handling complex imperfect important Such particularly background. when with integration is scenes
Such integration is particularly important when handling complex scenes with imperfect background.
38,750
built. how the framework are individual is how and We the developed, components show
We show how the framework is developed, and how the individual components are built.
38,751
and experiments studies Extensive evaluate proposed are method. to conducted the ablation
Extensive experiments and ablation studies are conducted to evaluate the proposed method.
38,752
this paper, tackle framework new we to propose the above In problems. a
In this paper, we propose a new framework to tackle the above problems.
38,753
MTV B-spline registration characterize the to shrinkage. used large regularized diffeomorphic was
B-spline regularized diffeomorphic registration was used to characterize the large MTV shrinkage.
38,754
MTV measure Jacobian changes. to the transformation of local was resulting the computed
Jacobian of the resulting transformation was computed to measure the local MTV changes.
38,755
using to HAR on focuses research due smartphones use. Recent widespread their on
Recent research on HAR focuses on using smartphones due to their widespread use.
38,756
can training online first inference. both the framework that This paper presents HAR and perform
This paper presents the first HAR framework that can perform both online training and inference.
38,757
broad deal vision must challenging an extremely of vehicle with set and scenes. Self-driving systems
Self-driving vehicle vision systems must deal with an extremely broad and challenging set of scenes.
38,758
high that This in labeled objects with images approach are assumes accuracy.
This approach assumes that objects in images are labeled with high accuracy.
38,759
generate vehicle in video. of coherence labels the accurate we exploit the field, To spatio-temporal
To generate accurate labels in the field, we exploit the spatio-temporal coherence of vehicle video.
38,760
of several the we feasibility paper In steps. approach this in demonstrate this
In this paper we demonstrate the feasibility of this approach in several steps.
38,761
we describe self-labeling scheme a using tracking. object Finally,
Finally, we describe a self-labeling scheme using object tracking.
38,762
complexity generate and geometry Evolution similar in ways.
Evolution and geometry generate complexity in similar ways.
38,763
The order. shape successively random pairs in were presented
The shape pairs were presented successively in random order.
38,764
joint (HMC) within Monte Bayesian We Gibbs using Hamiltonian sampling. develop Carlo inference
We develop joint Bayesian inference using Hamiltonian Monte Carlo (HMC) within Gibbs sampling.
38,765
hormones plant cardinal considered in Auxin is growth of and one development. the
Auxin is considered one of the cardinal hormones in plant growth and development.
38,766
throughout of the a plant. range It processes wide regulates
It regulates a wide range of processes throughout the plant.
38,767
pathway agrochemicals. valuable the Synthetic are auxin-signalling and herbicidal auxins exploit as
Synthetic auxins exploit the auxin-signalling pathway and are valuable as herbicidal agrochemicals.
38,768
work auxin of of chemistry for the auxins. design receptor-selective This expands the range synthetic
This work expands the range of auxin chemistry for the design of receptor-selective synthetic auxins.
38,769
Logic received non-monotonic interest. In years, Programming Inductive growing recent has
In recent years, non-monotonic Inductive Logic Programming has received growing interest.
38,770
noise-tolerant of learning from we present a the In framework. paper, this sets generalisation answer
In this paper, we present a noise-tolerant generalisation of the learning from answer sets framework.
38,771
AG models source the is available. code proposed publicly for The
The source code for the proposed AG models is publicly available.
38,772
is of technology Artificial and transformation. the technological intelligence (AI) core industrial revolution
Artificial intelligence (AI) is the core technology of technological revolution and industrial transformation.
38,773
work, this survey intelligence. financial we In on studies existing
In this work, we survey existing studies on financial intelligence.
38,774
with much spatial fewer effective CapsNet parameters. has been proved modeling features in
CapsNet has been proved effective in modeling spatial features with much fewer parameters.
38,775
number is misery found The of has routing be to optimal which procedure manually.
The optimal number of routing procedure is misery which has to be found manually.
38,776
implement We "convolutional" of version "full-connected" and both G-CapsNet of version G-CapsNet.
We implement both "full-connected" version of G-CapsNet and "convolutional" version of G-CapsNet.
38,777
achieves similar MNIST the as papers. dataset a G-CapsNet performance in the in original
G-CapsNet achieves a similar performance in the dataset MNIST as in the original papers.
38,778
layers. capsule stacking possibility we multiple of also Besides, explored
Besides, we also explored possibility of stacking multiple capsule layers.
38,779
on modular detailed a ontology We ontology patterns. based modeling for provide example design
We provide a detailed example for modular ontology modeling based on ontology design patterns.
38,780
that results outperforms the Experimental method methods. proposed conventional reveal demosaicking
Experimental results reveal that the proposed method outperforms conventional demosaicking methods.
38,781
get feature better employing representations by learn pyramids. usually multi-scale object detectors State-of-the-art results to
State-of-the-art object detectors usually learn multi-scale representations to get better results by employing feature pyramids.
38,782
in-place, end-to-end reconfiguration and the local global attention lightweight, and Both trainable. are
Both the global attention and local reconfiguration are lightweight, in-place, and end-to-end trainable.
38,783
using and build learning detect classify model heart deep a We disease to $X-ray$.
We build a deep learning model to detect and classify heart disease using $X-ray$.
38,784
datasets. data We several from collect and hospitals public
We collect data from several hospitals and public datasets.
38,785
management current conservation. tourism biodiversity achieving approaches Our are not
Our current tourism management approaches are not achieving biodiversity conservation.
38,786
insulin and sensitivity. of Peroxisome gamma proliferator-activated controllers ligand-activated (PPAR{\gamma}) metabolic are various actions receptors
Peroxisome proliferator-activated receptors gamma (PPAR{\gamma}) are ligand-activated controllers of various metabolic actions and insulin sensitivity.
38,787
is (S-AOG) indoor And-Or to spatial attributed An graph proposed scenes. represent
An attributed spatial And-Or graph (S-AOG) is proposed to represent indoor scenes.
38,788
in are nodes which grammar the a The terminal S-AOG model, entities. is object probabilistic
The S-AOG is a probabilistic grammar model, in which the terminal nodes are object entities.
38,789
required. So, pixel pixel no is calculation by complex
So, no complex pixel by pixel calculation is required.
38,790
through different Potential tested is features medical dataset. and established of proposed
Potential of proposed features is tested and established through different medical dataset.
38,791
in and vision applications tasks. wide have Dense cues depth important computer various are
Dense depth cues are important and have wide applications in various computer vision tasks.
38,792
maps obtained hardware its sparse depth generally limitation. of However, by LIDAR are because
However, depth maps obtained by LIDAR are generally sparse because of its hardware limitation.
38,793
depth incorporated the improve could be to performance. RGB features further completion Additional
Additional RGB features could be incorporated to further improve the depth completion performance.
38,794
there growing for to is a moving automatically need Accordingly, techniques images. classify
Accordingly, there is a growing need for techniques to automatically classify moving images.
38,795
processes hybridization. Phylogenetic such networks for to trees modeling allow as phylogenetic reticulate extend evolutionary
Phylogenetic networks extend phylogenetic trees to allow for modeling reticulate evolutionary processes such as hybridization.
38,796
parallel in discuss structures. population the genetics efforts inferring community at aimed similar also We
We also discuss parallel efforts in the population genetics community aimed at inferring similar structures.
38,797
we research Finally, highlight major for directions area. in future this
Finally, we highlight major directions for future research in this area.
38,798
and important recently extensively. investigated issue is an been has removal Rain streak
Rain streak removal is an important issue and has recently been investigated extensively.
38,799