text
stringlengths
27
153
label
stringlengths
27
153
id
int64
0
40k
parameters/variables modes responsible of We identified oscillation. for the different key
We identified the key parameters/variables responsible for different modes of oscillation.
15,200
modes flexible in the between appropriately parameters/variables. by network The choosing switching is required different
The network is flexible in switching between different modes by choosing appropriately the required parameters/variables.
15,201
Third, greyscale useful the using Un-sharp we in Masking method. image sharpen information resulting the
Third, we sharpen the useful information in the resulting greyscale image using Un-sharp Masking method.
15,202
accuracy. can optical recognition significantly The and detection character improve approach text rate proposed
The proposed approach can significantly improve text detection rate and optical character recognition accuracy.
15,203
depend may in on mutant probability arises. structured the where However, fixation a this population,
However, in a structured population, this fixation probability may depend on where the mutant arises.
15,204
epidemic only decisions information. algorithms Traditional detection using make local
Traditional epidemic detection algorithms make decisions using only local information.
15,205
several explicitly fusion metapopulations. novel We a spatial from information that models propose approach
We propose a novel approach that explicitly models spatial information fusion from several metapopulations.
15,206
regarding takes the considerations also announcement into cost-benefit method epidemic. Our account of
Our method also takes into account cost-benefit considerations regarding the announcement of epidemic.
15,207
adaptive, The alarms. false of strategy non-parametric detection probability detection balances delay vis-a-vis resulting optimally
The resulting adaptive, non-parametric detection strategy optimally balances detection delay vis-a-vis probability of false alarms.
15,208
to proper predictions. model a crucial making is accurate time in characterize a Using series
Using a proper model to characterize a time series is crucial in making accurate predictions.
15,209
demonstrating results performance are of approach. proposed Numerical provided the the
Numerical results are provided demonstrating the performance of the proposed approach.
15,210
lot a recognition attention during last decades. drawn of has Fingerprint
Fingerprint recognition has drawn a lot of attention during last decades.
15,211
and have used recognition past. fingerprint been algorithms for features the Different in
Different features and algorithms have been used for fingerprint recognition in the past.
15,212
representation scattering paper, In transform/network, for powerful used called recognition. image a is this
In this paper, a powerful image representation called scattering transform/network, is used for recognition.
15,213
predefined network are Scattering transforms. its where network is wavelet filters and a architecture convolutional
Scattering network is a convolutional network where its architecture and filters are predefined wavelet transforms.
15,214
of After principal extraction (PCA). analysis is features, reduced component their applying dimensionality scattering by
After extraction of scattering features, their dimensionality is reduced by applying principal component analysis (PCA).
15,215
matching task. template used SVM for perform the end, is At recognition multi-class to the
At the end, multi-class SVM is used to perform template matching for the recognition task.
15,216
However, are optimal used in solutions on functions. dependent the fundamentally objective parameters the the
However, the optimal solutions are fundamentally dependent on the parameters used in the objective functions.
15,217
range. maps their highly colour over provided perceptual Many vendors have by contrast uneven
Many colour maps provided by vendors have highly uneven perceptual contrast over their range.
15,218
also maps may Colour features. the perceptual that false have appearance induce of discontinuities
Colour maps may also have perceptual discontinuities that induce the appearance of false features.
15,219
specific are maps rainbow and developed The cyclic colour linear, detail. for requirements diverging, in
The specific requirements for linear, diverging, rainbow and cyclic colour maps are developed in detail.
15,220
images this To for colour maps support evaluating work presented. are two test
To support this work two test images for evaluating colour maps are presented.
15,221
construction of Finally, set of basis images colours the new presented. ternary for a are
Finally, a set of new basis colours for the construction of ternary images are presented.
15,222
simple programs produce many can models variables. Even probabilistic infinitely with
Even simple probabilistic programs can produce models with infinitely many variables.
15,223
classification. Existing their on convolutional shown success deep great have networks image (CNNs) neural
Existing deep convolutional neural networks (CNNs) have shown their great success on image classification.
15,224
such However, representation. explicit image important very generating are for dependencies
However, such dependencies are very important for generating explicit image representation.
15,225
RNNs can be applied directly hardly General non-sequential on data.
General RNNs can hardly be directly applied on non-sequential data.
15,226
we proposed the Thus, hierarchical RNNs (HRNNs).
Thus, we proposed the hierarchical RNNs (HRNNs).
15,227
in respectively. it all argument an is accepted, one extensions, or is if eventually accepted
an argument is eventually accepted, if it is accepted in one or all extensions, respectively.
15,228
by presented. semantics are axiomatic ranking on semantics our counting Lastly, induced an the perspective
Lastly, an axiomatic perspective on the ranking semantics induced by our counting semantics are presented.
15,229
Target Tracking Multiple is often tracking-by-detection within addressed the Online paradigm. (MTT)
Online Multiple Target Tracking (MTT) is often addressed within the tracking-by-detection paradigm.
15,230
ambiguities or arise of errors. in Nevertheless, presence occlusions detection
Nevertheless, ambiguities arise in presence of occlusions or detection errors.
15,231
(e.g., "Why questions hands?") we test shake do The
The test questions (e.g., "Why do we shake hands?")
15,232
a based used on The answering ConceptNet of version methods. spectral question
The question answering used a version of ConceptNet based on spectral methods.
15,233
Large indicate areas potential subtests improvement. of among variations
Large variations among subtests indicate potential areas of improvement.
15,234
sense. common strongly most is Comprehension subtest associated the with
Comprehension is the subtest most strongly associated with common sense.
15,235
on research AI. work continues Psychometric this Also, previous
Also, this work continues previous research on Psychometric AI.
15,236
Parts an Expanded (e.g. Model for (EPM) recognizing We introduce attributes human
We introduce an Expanded Parts Model (EPM) for recognizing human attributes (e.g.
15,237
suit) actions short young, (e.g. wearing and hair,
young, short hair, wearing suit) and actions (e.g.
15,238
consist is few to models contrast of (i.e. a in relatively which current This
This is in contrast to current models which consist of a relatively few (i.e.
15,239
in image. an random it redundant ignores background and
it ignores redundant and random background in an image.
15,240
attributes human and actions. our validate on We of three datasets recent method challenging
We validate our method on three recent challenging datasets of human attributes and actions.
15,241
results convincing quantitative datasets. obtain qualitative on the three We and state-of-the-art
We obtain convincing qualitative and state-of-the-art quantitative results on the three datasets.
15,242
different faces, in etc.) persons major clothing, everyday Recognising challenges presents (occluded locations, photos
Recognising persons in everyday photos presents major challenges (occluded faces, different clothing, locations, etc.)
15,243
ones. existing challenging we limitations more the addition, propose benchmarks discuss of and In
In addition, we discuss the limitations of existing benchmarks and propose more challenging ones.
15,244
results based a architecture. clustered We model on these spiking using a network explain
We explain these results using a spiking network model based on a clustered architecture.
15,245
on an the predicts the Moreover, model of dimensionality. bound existence upper
Moreover, the model predicts the existence of an upper bound on dimensionality.
15,246
on and the trials. bounds empirical The depends of of duration estimation such number
The empirical estimation of such bounds depends on the number and duration of trials.
15,247
The unknown. mechanism these observations common underlying remains
The common mechanism underlying these observations remains unknown.
15,248
use of livers. then the otherwise We unobservable and metabolic model to properties quantify determine
We then use the model to determine and quantify otherwise unobservable metabolic properties of livers.
15,249
of a Clustering parallel Linear Iterative We (SLIC) segmentation. Simple introduce the superpixel GPU implementation
We introduce a parallel GPU implementation of the Simple Linear Iterative Clustering (SLIC) superpixel segmentation.
15,250
solving for useful Learning representations an unsupervised a instrument sparse learning feature is problem.
Learning sparse feature representations is a useful instrument for solving an unsupervised learning problem.
15,251
present n-MNIST. In datasets, handwritten paper, three this digit collectively called we labeled
In this paper, we present three labeled handwritten digit datasets, collectively called n-MNIST.
15,252
already formalized New often proof concepts developments similar to involve assistant ones.
New proof assistant developments often involve concepts similar to already formalized ones.
15,253
can combined advice. be internal external with further help Such
Such external help can be further combined with internal advice.
15,254
dimensional high in are data. successful methods handling scale large based Vector quantization
Vector quantization based methods are successful in handling large scale high dimensional data.
15,255
computing perform methods datasets, distance excessive computations. these large However for
However for large datasets, these distance computing methods perform excessive computations.
15,256
also cost. lower to Encoding We further the Forest(E-Forest) computation propose
We also propose Encoding Forest(E-Forest) to further lower the computation cost.
15,257
quantization-based compatible is with E-Tree E-Forest methods. existing and various
E-Tree and E-Forest is compatible with various existing quantization-based methods.
15,258
scale perform approximate Quantization large methods been to search tasks. introduced have nearest
Quantization methods have been introduced to perform large scale approximate nearest search tasks.
15,259
quantization of one Vector Residual is (RVQ) the Quantization methods. effective
Residual Vector Quantization (RVQ) is one of the effective quantization methods.
15,260
quantization by codebook multi-stage the scheme stage a to stage. RVQ error learning uses lower
RVQ uses a multi-stage codebook learning scheme to lower the quantization error stage by stage.
15,261
performance quickly with The added gain diminishes stages.
The performance gain diminishes quickly with added stages.
15,262
vector actually NP-hard. Encoding a RVQ with is
Encoding a vector with RVQ is actually NP-hard.
15,263
a human presents wearing paper faces persons of This database spectacles. for
This paper presents a database of human faces for persons wearing spectacles.
15,264
to the valuable As be community. a can contribution this computer such, vision
As such, this can be a valuable contribution to the computer vision community.
15,265
This note Jabbari al. presented by shows that the et Nooghabi results
This note shows that the results presented by Jabbari Nooghabi et al.
15,266
with analysis. proposed the framework support $(K,\epsilon)$-robustness We further a formal
We further support the proposed framework with a formal $(K,\epsilon)$-robustness analysis.
15,267
periodically to monitored tracks safe transportation. ensure inspected and need Railroad to be
Railroad tracks need to be periodically inspected and monitored to ensure safe transportation.
15,268
indispensable is Fine-tuned cellular replication nucleotide for pools the regulation DNA. of of faithful
Fine-tuned regulation of the cellular nucleotide pools is indispensable for faithful replication of DNA.
15,269
DNA also damage information The recognition is repair genetic and by processes. safeguarded
The genetic information is also safeguarded by DNA damage recognition and repair processes.
15,270
fast effective to encodings idea The allow generate to is approximation. distance
The idea is to generate effective encodings to allow fast distance approximation.
15,271
than quantization is state-of-the-art. The error other lower
The quantization error is lower than other state-of-the-art.
15,272
search a efficient non-exhaustive HCLAE ANN-Search. (A-Tree), propose Aggregating-Tree we using perform to method Further,
Further, we propose Aggregating-Tree (A-Tree), a non-exhaustive search method using HCLAE to perform efficient ANN-Search.
15,273
state-of-the-art. magnitudes on ANN-Search to the achieves of compared speed-up tasks, A-Tree
A-Tree achieves magnitudes of speed-up on ANN-Search tasks, compared to the state-of-the-art.
15,274
noise protein duplication of influences different how systematically genome components. the We investigate timing
We systematically investigate how the timing of genome duplication influences different protein noise components.
15,275
limitations. a this is these overcome paper, proposed new of version In EFIS to
In this paper, a new version of EFIS is proposed to overcome these limitations.
15,276
work of focus image generation. on we In problem caption the this
In this work we focus on the problem of image caption generation.
15,277
Bayesian flexible Macau, factorization propose a data. and powerful for heterogeneous method We
We propose Macau, a powerful and flexible Bayesian factorization method for heterogeneous data.
15,278
the Santos a of The is and modification algorithm Method. Proenka
The algorithm is a modification of the Santos and Proenka Method.
15,279
the the region. segment detected from We sclera, then eye out
We then segment out the sclera, from the detected eye region.
15,280
contour. segmented an sclera, the approximate segment we eyelid out From
From the segmented sclera, we segment out an approximate eyelid contour.
15,281
corner candidates detector. are and Stephens obtained Eye corner Harris using
Eye corner candidates are obtained using Harris and Stephens corner detector.
15,282
eye the locate finally. the corners, post-pruning corner candidates a to of Eye We introduce
We introduce a post-pruning of the Eye corner candidates to locate the eye corners, finally.
15,283
as on database. well created been has The Yale, as algorithm databases tested our JAFFE
The algorithm has been tested on Yale, JAFFE databases as well as our created database.
15,284
human We scheme facial evaluate landmarks and proposed localisation. the sheep both on
We evaluate the proposed scheme on both human and sheep facial landmarks localisation.
15,285
for of approach an characteristics this work, semantic-part we present representations. object analyzing In category
In this work, we present an approach for analyzing semantic-part characteristics of object category representations.
15,286
For experiments, for representation we sketch-based objects. recently use spatial our a category-epitome, proposed
For our experiments, we use category-epitome, a recently proposed sketch-based spatial representation for objects.
15,287
convolutional a perform object How network (FCN) can detection? fully neural single on
How can a single fully convolutional neural network (FCN) perform on object detection?
15,288
proposed We et. network spatial [Jaderberg the recently (SPN) transformer integrate
We integrate the recently proposed spatial transformer network (SPN) [Jaderberg et.
15,289
digits classify to MNIST RNN-SPN sequences. in We cluttered use the
We use the RNN-SPN to classify digits in cluttered MNIST sequences.
15,290
a version SPN image. The skewed zoomed, the and outputs input rotated of
The SPN outputs a zoomed, rotated and skewed version of the input image.
15,291
capacity. to Numerical artificial lighting extend tool have introduced the been simulation of codes
Numerical simulation codes of artificial lighting have been introduced to extend the tool capacity.
15,292
for original these model calculations The semi-detailed (simplified). and used is
The model used for these calculations is original and semi-detailed (simplified).
15,293
having reliability a of idea the of numerical precise This about simulations. results allowed the
This allowed having a precise idea about the reliability of the results of numerical simulations.
15,294
non-smooth sparsity allow parameter We for in non-convex to penalties induce regularisation weights.
We allow for non-smooth non-convex regularisation penalties to induce sparsity in parameter weights.
15,295
apply focus generally. We deep learning on logistic more although our methods multiclass
We focus on deep multiclass logistic learning although our methods apply more generally.
15,296
with we research. Finally, for future directions conclude
Finally, we conclude with directions for future research.
15,297
image-set To approach. this, querying an propose based localization we address
To address this, we propose an image-set querying based localization approach.
15,298
proposed and shown effectiveness feasibility have Experiments approach. the of the
Experiments have shown the effectiveness and feasibility of the proposed approach.
15,299