text
stringlengths
27
153
label
stringlengths
27
153
id
int64
0
40k
We benchmark performance state-of-the-art datasets. in achieve extensive
We achieve state-of-the-art performance in extensive benchmark datasets.
39,000
iris recognition. of paper post-mortem This analysis presents unique a human
This paper presents a unique analysis of post-mortem human iris recognition.
39,001
tracking a In using a optimally we as approach. practice, path problem solve K-shortest this
In practice, we solve this optimally as a tracking problem using a K-shortest path approach.
39,002
these are of relation Insightful all conclusions to aspects. offered three in
Insightful conclusions are offered in relation to all three of these aspects.
39,003
Thus, has size departments. railway from increasing EMU associated fleet the attention attracted prediction of
Thus, the prediction of EMU fleet size has attracted increasing attention from associated railway departments.
39,004
in two networks analyzed differences of computation The and performances are these experiments in-depth.
The differences and performances of these two networks in computation experiments are analyzed in-depth.
39,005
of is object nearby an among symmetries and among Distinction poses challenging.
Distinction among nearby poses and among symmetries of an object is challenging.
39,006
In we group-theoretic both. a propose unified, this to paper, approach tackle
In this paper, we propose a unified, group-theoretic approach to tackle both.
39,007
to We matrices. sample Metropolis-Hastings from algorithm space propose the correlation of uniformly a novel
We propose a novel Metropolis-Hastings algorithm to sample uniformly from the space of correlation matrices.
39,008
no from However or other populations. population fully animal human human is isolated
However no human population is fully isolated from other human or animal populations.
39,009
neural improvements deep object by to have Driven pose brought networks, methods recent estimation. significant
Driven by deep neural networks, recent methods have brought significant improvements to object pose estimation.
39,010
the have approach. encourage research, To we reproducible proposed for released codes our
To encourage reproducible research, we have released the codes for our proposed approach.
39,011
features. Convolutional starting learning neural networks are local to able with hierarchical process perform a
Convolutional neural networks are able to perform a hierarchical learning process starting with local features.
39,012
features However, attention paid a to is like limited level elementary enhancing edges. such
However, a limited attention is paid to enhancing such elementary level features like edges.
39,013
transmission. of misclassification community of community-onset role underplays the and cases Underreporting systematic
Underreporting of community-onset cases and systematic misclassification underplays the role of community transmission.
39,014
Harmonic, by Heronian Geometric, means have been Arithmetic, and Contraharmonic mathematicians. many studied
Harmonic, Geometric, Arithmetic, Heronian and Contraharmonic means have been studied by many mathematicians.
39,015
on image superior Convolutional performance recent Neural depth single have in Networks years. estimation demonstrated
Convolutional Neural Networks have demonstrated superior performance on single image depth estimation in recent years.
39,016
The combines dilated ResNet and DFE convolutions.
The DFE combines ResNet and dilated convolutions.
39,017
input from multi-scale extracts keeping while information maps feature image It dense. the
It extracts multi-scale information from input image while keeping the feature maps dense.
39,018
attention DMG, we features multi-scale DFE. As to for in mechanism fuse use produced
As for DMG, we use attention mechanism to fuse multi-scale features produced in DFE.
39,019
is any does Network need Our and trained post-processing. not end-to-end
Our Network is trained end-to-end and does not need any post-processing.
39,020
this an domain brings option. appealing adaptation setting, In
In this setting, domain adaptation brings an appealing option.
39,021
give Conservative studies Ablation the insights more properties into Loss. of
Ablation studies give more insights into properties of the Conservative Loss.
39,022
method a images introduce learning perturbations for adversarial to or individual targeted We videos.
We introduce a method for learning adversarial perturbations targeted to individual images or videos.
39,023
data This discussed in prospective of augmentation terms a is scheme.
This is discussed in terms of a prospective data augmentation scheme.
39,024
for yet high-quality perturbations also sparse The or be video leveraged compression. image may
The sparse yet high-quality perturbations may also be leveraged for image or video compression.
39,025
this generalizing learn discriminant The to views. representations across is view-invariant to well key problem
The key to this problem is to learn discriminant view-invariant representations generalizing well across views.
39,026
on results that datasets the show outperforms four Experimental our approach state-ofthe-art multiview approaches.
Experimental results on four multiview datasets show that our approach outperforms the state-ofthe-art approaches.
39,027
and detail procedure. in paper evaluation the This presents dataset
This paper presents in detail the dataset and evaluation procedure.
39,028
The with for directions paper concludes improvements. future
The paper concludes with directions for future improvements.
39,029
can Sensing scenario during medical the the safety operations. ensure the surgical
Sensing the medical scenario can ensure the safety during the surgical operations.
39,030
follows. work contributions as our of are The key
The key contributions of our work are as follows.
39,031
Systems have convolutional perform remarkable manipulation achieved realism. networks deep image that using
Systems that perform image manipulation using deep convolutional networks have achieved remarkable realism.
39,032
despite applications pose large require changing precise Many environments. and estimates operating in robotics
Many robotics applications require precise pose estimates despite operating in large and changing environments.
39,033
biology Diffusion important is cell and in complicated.
Diffusion in cell biology is important and complicated.
39,034
to additional as leverage signals. consistency geometric supervisory In propose we this paper,
In this paper, we propose to leverage geometric consistency as additional supervisory signals.
39,035
favorably methods. that our compare unsupervised and depth state-of-the-art Extensive demonstrate flow experiments with models
Extensive experiments demonstrate that our depth and flow models compare favorably with state-of-the-art unsupervised methods.
39,036
It this roadmap issue. tackle general presented a to
It presented a general roadmap to tackle this issue.
39,037
well The samples the of when cGAN even works training are limited. number
The cGAN works well even when the number of training samples are limited.
39,038
Therefore, the approaches. several outperforms proposed method state-of-the-art
Therefore, the proposed method outperforms several state-of-the-art approaches.
39,039
shown capability intelligence artificial tasks. various of Convolutional networks (CNNs) have solving great neural
Convolutional neural networks (CNNs) have shown great capability of solving various artificial intelligence tasks.
39,040
in the However, applications. raised in resource-limited has model challenges employing them increasing size
However, the increasing model size has raised challenges in employing them in resource-limited applications.
39,041
this Based light-weight operation, we ChannelNets. CNNs build known as novel on
Based on this novel operation, we build light-weight CNNs known as ChannelNets.
39,042
fibrillation. and arrhythmias dangerous synchronisation underlie fatal regime of abnormal self-organised Cardiac re-entry
Cardiac re-entry regime of self-organised abnormal synchronisation underlie dangerous arrhythmias and fatal fibrillation.
39,043
models Re-entry cardiac prescribed of the is locations mono-domain spatially tissue. initiated homogeneous at in
Re-entry is initiated at prescribed locations in the spatially homogeneous mono-domain models of cardiac tissue.
39,044
and consequently, they some to However, occur are automatically. traffic rarely recognize difficult signs
However, some traffic signs occur rarely and consequently, they are difficult to recognize automatically.
39,045
keypoints has in presence the an detecting be of unresolved to blur remained issue. However,
However, detecting keypoints in the presence of blur has remained to be an unresolved issue.
39,046
limited performance. a result, have those methods As
As a result, those methods have limited performance.
39,047
make scale-robust, it also to scale it To space. extend we
To make it scale-robust, we also extend it to scale space.
39,048
acceptance.} the will any our publicly our Without parallelization, upon available make code implementation\footnote{We
Without any parallelization, our implementation\footnote{We will make our code publicly available upon the acceptance.}
39,049
of results process, co-decision by acquired the classification learners. testing are During two
During testing process, classification results are acquired by co-decision of the two learners.
39,050
However, is label an process. expensive and time-consuming collection
However, label collection is an expensive and time-consuming process.
39,051
addition, view-adversarial propose In to training method view-invariant a we of learning features. enhance
In addition, we propose a view-adversarial training method to enhance learning of view-invariant features.
39,052
of recognition the We representations datasets. on demonstrate the action learned effectiveness for multiple
We demonstrate the effectiveness of the learned representations for action recognition on multiple datasets.
39,053
method we generic, images. segmentation blood vessels our for Though apply of in retinal fundus
Though generic, we apply our method for segmentation of blood vessels in retinal fundus images.
39,054
extraction. framework, for architecture CNN the spectrograms single In is a onto applied feature multiple
In the framework, a single CNN architecture is applied onto multiple spectrograms for feature extraction.
39,055
then fused spectrograms The acoustic extracted from features the to scenes. multiple discriminate are deep
The deep features extracted from multiple spectrograms are then fused to discriminate the acoustic scenes.
39,056
Experimental accuracies proposed that achieve show the on can results method both promising datasets.
Experimental results show that the proposed method can achieve promising accuracies on both datasets.
39,057
structures enhanced can introduce into acid functionalities and base Incorporating metal-mediated stabilities. pairs nucleic new
Incorporating metal-mediated base pairs into nucleic acid structures can introduce new functionalities and enhanced stabilities.
39,058
G-duplexes are linear rigid B-DNA. more IMS than indicates that and
IMS indicates that G-duplexes are linear and more rigid than B-DNA.
39,059
used to with propose the experiments. were IMS compatible structures calculations DFT
DFT calculations were used to propose structures compatible with the IMS experiments.
39,060
and Extensive performance experiments of demonstrate MDCN models. the the over effectiveness state-of-the-art superior
Extensive experiments demonstrate the effectiveness and superior performance of MDCN over the state-of-the-art models.
39,061
the solved CCMDP, The in problems be so approach states efficiently. decouples can large that
The approach decouples states in the CCMDP, so that large problems can be solved efficiently.
39,062
estimators empirical assessed of proposed The an usefulness the in example. is
The usefulness of the proposed estimators is assessed in an empirical example.
39,063
of is important and images most one operations editing videos. Compositing for the
Compositing is one of the most important editing operations for images and videos.
39,064
The called of realism is composite of harmonization. process the improving often results
The process of improving the realism of composite results is often called harmonization.
39,065
mainly Previous images. on focus harmonization for approaches
Previous approaches for harmonization mainly focus on images.
39,066
this attack In work, step to the of harmonization. further we video one take problem
In this work, we take one step further to attack the problem of video harmonization.
39,067
well show on that composite dataset. synthetic the training our to dataset Experiments real-world generalizes
Experiments show that training on our synthetic dataset generalizes well to the real-world composite dataset.
39,068
and significant pose motion, challenges. However, date, occlusions motion fast to blur
However, to date, fast motion, motion blur and occlusions pose significant challenges.
39,069
to for features depth. represent hint these may predict important Hence, learning an
Hence, these features may represent an important hint for learning to predict depth.
39,070
these blur we depth-prediction studies, out-of-focus that improves From performances. greatly the show network
From these studies, we show that out-of-focus blur greatly improves the depth-prediction network performances.
39,071
visual Contrast information is in processing. crucial a factor
Contrast is a crucial factor in visual information processing.
39,072
directions of provide topic. in this emerging future research further possible We
We further provide possible directions of future research in this emerging topic.
39,073
intrinsic with we information an model within-person its identity a variations. face First, and representation
First, we model a face representation with an intrinsic identity information and its within-person variations.
39,074
paper access This (MEC) non-orthogonal investigates uplink multiple mobile-edge an network. (NOMA)-based computing
This paper investigates an uplink non-orthogonal multiple access (NOMA)-based mobile-edge computing (MEC) network.
39,075
shown be optimal. algorithm The is proposed successfully globally to
The proposed algorithm is successfully shown to be globally optimal.
39,076
the results better than show performance methods. algorithm proposed achieves Numerical that the conventional
Numerical results show that the proposed algorithm achieves better performance than the conventional methods.
39,077
communication communication the integrated The wireless is in generation networks. promising next radar and system
The integrated radar and communication system is promising in the next generation wireless communication networks.
39,078
energy. However, by confined its performance is limited the
However, its performance is confined by the limited energy.
39,079
proposed. wireless it, and to radar overcome communication system powered order In integrated a is
In order to overcome it, a wireless powered integrated radar and communication system is proposed.
39,080
problem subject minimization performances. constraints formulated radar to the and on is communication An energy
An energy minimization problem is formulated subject to constraints on the radar and communication performances.
39,081
optimized energy energy. jointly beamforming to The waveform are radar-communication the minimize consumption and
The energy beamforming and radar-communication waveform are jointly optimized to minimize the consumption energy.
39,082
solved challenging variable relaxation by is using non-convex The and problem auxiliary methods. semidefinite
The challenging non-convex problem is solved by using semidefinite relaxation and auxiliary variable methods.
39,083
It is obtained. proved solution optimal that be can the
It is proved that the optimal solution can be obtained.
39,084
proposed results Simulation our benchmark scheme. outperforms optimal the design demonstrate that
Simulation results demonstrate that our proposed optimal design outperforms the benchmark scheme.
39,085
lesion results in skin models automated show remarkable analysis. learning Deep
Deep learning models show remarkable results in automated skin lesion analysis.
39,086
input expand transforming by dataset training Data images. can the augmentation
Data augmentation can expand the training dataset by transforming input images.
39,087
GO-CNN on stability. amount reduces and examples enhances the Furthermore, adversarial dependence of training
Furthermore, GO-CNN reduces dependence on the amount of training examples and enhances adversarial stability.
39,088
addresses paper optimal sampling the set selecting for problem on graphs. This an of signals
This paper addresses the problem of selecting an optimal sampling set for signals on graphs.
39,089
through execution our the We performance errors approach evaluate time. of and prediction comparisons of
We evaluate the performance of our approach through comparisons of prediction errors and execution time.
39,090
this from suffers still great limitations. framework success, several Despite the tracking
Despite the great success, this tracking framework still suffers from several limitations.
39,091
it object large cannot First, properly handle rotation.
First, it cannot properly handle large object rotation.
39,092
distracted Second, gets salient the background objects. contains tracking when easily
Second, tracking gets easily distracted when the background contains salient objects.
39,093
Increasing existing thumbnail capabilities direction. compression research codecs of an beyond is therefore the active
Increasing thumbnail compression beyond the capabilities of existing codecs is therefore an active research direction.
39,094
triangles. present we the algorithm for Second, novel encoding a
Second, we present a novel algorithm for encoding the triangles.
39,095
identifying by for promising research. future survey We finish the directions
We finish the survey by identifying promising directions for future research.
39,096
has transfer learning acquired interest. recently Deep research significant
Deep transfer learning recently has acquired significant research interest.
39,097
deep Model-based probably used most transfer the is method. frequently learning
Model-based deep transfer learning is probably the most frequently used method.
39,098
approach, learning features. useful help learning more Using can transfer models to deep capture this
Using this approach, transfer learning can help deep learning models to capture more useful features.
39,099