text
stringlengths
27
153
label
stringlengths
27
153
id
int64
0
40k
Treatment measurements contemplate some protocols usually assess order to recurrent such as progress, in goniometry.
Treatment protocols usually contemplate some recurrent measurements in order to assess progress, such as goniometry.
37,600
to the angle. flexion are values These estimate sufficient
These values are sufficient to estimate the flexion angle.
37,601
were for coordinates. these trained predicting networks six neural Convolutional then
Convolutional neural networks were then trained for predicting these six coordinates.
37,602
van T. works contraction Kampen followed of Kurtz. Their of and the
Their contraction followed the works of van Kampen and of T. Kurtz.
37,603
connected variables S LE. This to the noise the in subunit channel terms the
This S connected the noise terms to the channel subunit variables in the LE.
37,604
that the of numerous Subsequently, were studies these models. simplified several showed shortcomings published
Subsequently, numerous studies were published that showed the several shortcomings of these simplified models.
37,605
that results it with produced good results. agreement the in They MC very showed
They showed that it produced results in very good agreement with the MC results.
37,606
regression shown classification CNN-based performance and significant improvements this for task. methods have Recently,
Recently, CNN-based regression and classification methods have shown significant performance improvements for this task.
37,607
allows instances. This pixels to object individual us group into
This allows us to group object pixels into individual instances.
37,608
Photoacoustic most beamformer. used In (DAS) commonly is the imaging, algorithm Delay-and-Sum
In Photoacoustic imaging, Delay-and-Sum (DAS) algorithm is the most commonly used beamformer.
37,609
and leads of low level sidelobes. a it high to resolution However,
However, it leads to a low resolution and high level of sidelobes.
37,610
sidelobes (DMAS) introduced compared to Delay-Multiply-and-Sum was DAS. to provide lower
Delay-Multiply-and-Sum (DMAS) was introduced to provide lower sidelobes compared to DAS.
37,611
artificially examples, of reducing increasing improves it and By the generalization. overfitting helps number training
By artificially increasing the number of training examples, it helps reducing overfitting and improves generalization.
37,612
actually Otherwise, strategy we the hurts. previous show that
Otherwise, we show that the previous strategy actually hurts.
37,613
to CT pixel parameters. differences have in due varying However, spacing acquisition studies
However, CT studies have varying pixel spacing due to differences in acquisition parameters.
37,614
improve the are performance. used ensembles further also Multi-model to
Multi-model ensembles are also used to further improve the performance.
37,615
of images. consider problem scene search We person the unconstrained in
We consider the problem of person search in unconstrained scene images.
37,616
multi-scale person barrier. is matching problem under-studied the in Faster-RCNN), severe a and more search
Faster-RCNN), and the under-studied multi-scale matching problem in person search is a more severe barrier.
37,617
to facilitate simultaneously increasing Automatic segmentation potential has while efficiency. morphological great measurements
Automatic segmentation has great potential to facilitate morphological measurements while simultaneously increasing efficiency.
37,618
closed is for expression CSCRB the derived. A form
A closed form expression for the CSCRB is derived.
37,619
such Up been image have approaches to limited single exclusively to now, however, super-resolution.
Up to now, however, such approaches have been exclusively limited to single image super-resolution.
37,620
recurrent proposed information architecture, previous from frames to due its leverages i.e. architecture naturally The
The proposed architecture naturally leverages information from previous frames due to its recurrent architecture, i.e.
37,621
provide a novel combining merging for (Eg. method We view information view dependent
We provide a novel view merging method for combining view dependent information (Eg.
37,622
paper new This learning video-based for a classification. deep approach scene presents
This paper presents a new deep learning approach for video-based scene classification.
37,623
reliable assist to methods doctor. required are and Automatic segmenta-tion
Automatic and reliable segmenta-tion methods are required to assist doctor.
37,624
is deep for ordinary. the inspiration, the always which rare improvement networks But of needs
But the improvement of deep networks always needs inspiration, which is rare for the ordinary.
37,625
These are own have different for data characteristic distinctive their methods and sets.
These different methods have their own characteristic and are distinctive for data sets.
37,626
structure. on dynamics network network Neuronal depends
Neuronal network dynamics depends on network structure.
37,627
processors for to known bias "burn-in" complicate the of use parallel computations. is This MCMC
This "burn-in" bias is known to complicate the use of parallel processors for MCMC computations.
37,628
across recognition robustly face wide variety Current recognize of identities conditions. systems a imaging
Current face recognition systems robustly recognize identities across a wide variety of imaging conditions.
37,629
or problem problem polycube construction deformation of The graphics. computer in is an essential
The problem of polycube construction or deformation is an essential problem in computer graphics.
37,630
corresponding its a shape. a derive polycube We between mesh relationship clear and
We derive a clear relationship between a mesh and its corresponding polycube shape.
37,631
with works meshes boundaries. Our without is algorithm or and on edge-preserved, surface
Our algorithm is edge-preserved, and works on surface meshes with or without boundaries.
37,632
in Our outperforms algorithm efficiency. ones speed, previous robustness,
Our algorithm outperforms previous ones in speed, robustness, efficiency.
37,633
methods. favorably other that compares We polycube demonstrat to method deformation our state-of-the-art
We demonstrat that our method compares favorably to other state-of-the-art polycube deformation methods.
37,634
longer often generalise RNNs training. the to sequences ones to However, than during struggle encountered
However, RNNs often struggle to generalise to sequences longer than the ones encountered during training.
37,635
induction. we In networks for to work, explicitly neural optimise propose this
In this work, we propose to optimise neural networks explicitly for induction.
37,636
brain brain with which generate models, compare functional models similar popular networks. our network We
We compare our brain network models with popular models, which generate similar functional brain networks.
37,637
and advertisements in non-literal interact text in complex, Images ways.
Images and text in advertisements interact in complex, non-literal ways.
37,638
Extensive conducted experiments effectiveness. were to its validate
Extensive experiments were conducted to validate its effectiveness.
37,639
identify classification to skin the disease. Melanoma stage is serious a
Melanoma classification is a serious stage to identify the skin disease.
37,640
cancer patients. biomarker important Tumor proliferation of indicative the of breast prognosis is an
Tumor proliferation is an important biomarker indicative of the prognosis of breast cancer patients.
37,641
a of in proliferation Assessment labor-intensive tumor subjective task. and highly is setting clinical
Assessment of tumor proliferation in a clinical setting is highly subjective and labor-intensive task.
37,642
investigated study that tumor This proliferation was assessment from WSIs. first the
This was the first study that investigated tumor proliferation assessment from WSIs.
37,643
variation and CT in large nodules in appearance suffer Lung size images.
Lung nodules suffer large variation in size and appearance in CT images.
37,644
training speed tackle propose a curriculum strategy To we general up problem, to training. this
To tackle this problem, we propose a general curriculum training strategy to speed up training.
37,645
we state-of-the-art outperforms that the In methods. network previous experiments, proposed demonstrate
In experiments, we demonstrate that the proposed network outperforms previous state-of-the-art methods.
37,646
queries suffers from the significant space latter expensive The and from former overhead.
The former suffers from expensive queries and the latter from significant space overhead.
37,647
hypothesis. supports the our and study of visualization The further learned context generalizability
The visualization and generalizability study of the learned context further supports our hypothesis.
37,648
energy a has succinct Calabi format. and explicit
Calabi energy has a succinct and explicit format.
37,649
Its under is conditions. and certain conformal flow convergent corresponding
Its corresponding flow is conformal and convergent under certain conditions.
37,650
this present RGB-D system RGB-iD namely SLAM we for a In SLAM. complete paper cameras,
In this paper we present a complete SLAM system for RGB-D cameras, namely RGB-iD SLAM.
37,651
keyframes the back-end receives which front-end. is from second thread The a
The second thread is a back-end which receives keyframes from the front-end.
37,652
propose RSIR (FCN). approach novel networks fully multi-label We with a therefore convolutional
We therefore propose a novel multi-label RSIR approach with fully convolutional networks (FCN).
37,653
ambiguous. from challenging self-occluded regions both and images the hidden is Inferring
Inferring the hidden regions from self-occluded images is both challenging and ambiguous.
37,654
propose two-pronged We these issues. a address approach to
We propose a two-pronged approach to address these issues.
37,655
achieved latent by novel probablistic a diversity view-specific learning space loss. with a is This
This is achieved by learning a probablistic latent space with a novel view-specific diversity loss.
37,656
qualitative of the significance performed quantitative highlight approach. the is Thorough to and analysis proposed
Thorough quantitative and qualitative analysis is performed to highlight the significance of the proposed approach.
37,657
alternating is non-convex developed minimization solve algorithm this, to this problem. Following an
Following this, an alternating minimization algorithm is developed to solve this non-convex problem.
37,658
that architecture proposed the volumetric encodes a labels knowledge the proportion. represent The
The proposed architecture encodes the knowledge that the labels represent a volumetric proportion.
37,659
with is designed custom A to learn loss intervals.
A custom loss is designed to learn with intervals.
37,660
achieved performance. AI have being level human few Despite target developers, several for a
Despite being a target for several AI developers, few have achieved human level performance.
37,661
surge communications. motivated is displays This VR/AR by of face-to-face recent hinder that
This is motivated by recent surge of VR/AR displays that hinder face-to-face communications.
37,662
was temporal two Video shot mechanisms. on based pooling representation
Video shot representation was based on two temporal pooling mechanisms.
37,663
been evaluation have Novel visualization protocols, and result strategies, accordingly. designed methods training
Novel training strategies, evaluation protocols, and result visualization methods have been designed accordingly.
37,664
However, into large does images. dependence the not dataset well on medical translate
However, the dependence on large dataset does not translate well into medical images.
37,665
course of model, the and this disappear inter-residue during time evolution. form the In contacts
In this model, inter-residue contacts form and disappear during the course of the time evolution.
37,666
differ geometrical model. substantially of that this properties We homopeptides the demonstrate various in
We demonstrate that the geometrical properties of various homopeptides differ substantially in this model.
37,667
two using of the introduction The adversarial single the problems network tackles discriminator. a
The introduction of the adversarial network tackles the two problems using a single discriminator.
37,668
kinds the paper of inconsistencies, this Adversarial address Macro-Micro the Net proposes two (MMAN). To
To address the two kinds of inconsistencies, this paper proposes the Macro-Micro Adversarial Net (MMAN).
37,669
dataset generalization our impressive pre-trained ability. model small a demonstrates PPSS, relatively On
On a relatively small dataset PPSS, our pre-trained model demonstrates impressive generalization ability.
37,670
is Cellular associated strain. contractility assumed imply often to an compressive
Cellular contractility is often assumed to imply an associated compressive strain.
37,671
This of stiffer for on substrates. mechanism a the molecules suggests mechanotransductive stretch-activation physical
This suggests a physical mechanism for the stretch-activation of mechanotransductive molecules on stiffer substrates.
37,672
on supporting adhesions, recent results focal whole-cell experimental mechanically-driven mechanotransduction.
focal adhesions, supporting recent experimental results on whole-cell mechanically-driven mechanotransduction.
37,673
non-Lambertian photometric problem This for paper of the surfaces. stereo addresses
This paper addresses the problem of photometric stereo for non-Lambertian surfaces.
37,674
on generalize we data, real datasets. synthetic can well Although PS-FCN train it on
Although we train PS-FCN on synthetic data, it can generalize well on real datasets.
37,675
This gap. to paper fill aims this
This paper aims to fill this gap.
37,676
nearly under conditions prove, same the with as We i.i.d. first
We first prove, under nearly the same conditions as with i.i.d.
37,677
data, convergence clustering. the weak under processes empirical multiway of
data, the weak convergence of empirical processes under multiway clustering.
37,678
consistency estimators, establish asymptotic various of that of including We variance et al. then Cameron
We then establish consistency of various asymptotic variance estimators, including that of Cameron et al.
37,679
network great Deep artificial applications. has neural intelligence achieved success in many convolution
Deep convolution neural network has achieved great success in many artificial intelligence applications.
37,680
large benchmark local for detectors. the We scale of present evaluation a feature
We present a large scale benchmark for the evaluation of local feature detectors.
37,681
the loss. signal supervision to classification Conventionally, as been penalize softmax trained with have CNNs
Conventionally, CNNs have been trained with softmax as supervision signal to penalize the classification loss.
37,682
deep considered face and recognition maximization features of ideal are minimization Such task. for
Such minimization and maximization of deep features are considered ideal for face recognition task.
37,683
shown results. to breast cancer in Deep mammograms recently approaches promising learning detection have
Deep learning approaches to breast cancer detection in mammograms have recently shown promising results.
37,684
generating we First, that high-resolution of synthetic capable mammogram show GANs are patches.
First, we show that GANs are capable of generating high-resolution synthetic mammogram patches.
37,685
performance. evaluate cancer classification breast augmented experimentally dataset the improve Next, we to using
Next, we experimentally evaluate using the augmented dataset to improve breast cancer classification performance.
37,686
arbitrary This captures paper over timestamps. describes network a that multimodal correlations
This paper describes a network that captures multimodal correlations over arbitrary timestamps.
37,687
two the The streams. approach fusion existing averages
The existing fusion approach averages the two streams.
37,688
fusion we propose for CNN. learning a a Shannon Here network pre-trained correlation with a
Here we propose a correlation network with a Shannon fusion for learning a pre-trained CNN.
37,689
complement methods. network to fusion approach existing This the found was
This approach was found to complement the existing network fusion methods.
37,690
the recognition of video multimodal the The results. accuracy correlation enhanced
The multimodal correlation enhanced the accuracy of the video recognition results.
37,691
constrained We propose algorithm policy objective function. a optimize to the gradient
We propose a policy gradient algorithm to optimize the constrained objective function.
37,692
applications. video many important Localization Temporal untrimmed Action (TAL) in for is
Temporal Action Localization (TAL) in untrimmed video is important for many applications.
37,693
generic, and carefully consider not special approaches scenes. remain of properties However, do the largely
However, approaches remain largely generic, and do not carefully consider the special properties of scenes.
37,694
are perform regions a then to used scene source recognition. features of as These
These regions are then used as a source of features to perform scene recognition.
37,695
pedestrian evaluated publicly on also Grand is available performance New the The Central database. York
The performance is also evaluated on the publicly available New York Grand Central pedestrian database.
37,696
that Producing but weren't. been answer over have seen images requires an could marginalizing
Producing an answer requires marginalizing over images that could have been seen but weren't.
37,697
This a as pooling}. descriptor for \emph{discriminative used the subspace is subspace dubbed video, then
This subspace is then used as a descriptor for the video, dubbed \emph{discriminative subspace pooling}.
37,698
video We and datasets experiments results. on several state-of-the-art demonstrate provide
We provide experiments on several video datasets and demonstrate state-of-the-art results.
37,699