Home << 1 2 3 4 5 6 7 8 9 10 >> [11–11]
List View
 | 
Citations
 | 
Details
   web
Records
Author G. Zahnd; Simone Balocco; A. Serusclat; P. Moulin; M. Orkisz; D. Vray
Title Progressive attenuation of the longitudinal kinetics in the common carotid artery: preliminary in vivo assessment Ultrasound in Medicine and Biology Type Journal Article
Year 2015 Publication Ultrasound in Medicine and Biology Abbreviated Journal UMB
Volume 41 Issue 1 Pages 339-345
Keywords Arterial stiffness; Atherosclerosis; Common carotid artery; Longitudinal kinetics; Motion tracking; Ultrasound imaging
Abstract Longitudinal kinetics (LOKI) of the arterial wall consists of the shearing motion of the intima-media complex over the adventitia layer in the direction parallel to the blood flow during the cardiac cycle. The aim of this study was to investigate the local variability of LOKI amplitude along the length of the vessel. By use of a previously validated motion-estimation framework, 35 in vivo longitudinal B-mode ultrasound cine loops of healthy common carotid arteries were analyzed. Results indicated that LOKI amplitude is progressively attenuated along the length of the artery, as it is larger in regions located on the proximal side of the image (i.e., toward the heart) and smaller in regions located on the distal side of the image (i.e., toward the head), with an average attenuation coefficient of -2.5 ± 2.0%/mm. Reported for the first time in this study, this phenomenon is likely to be of great importance in improving understanding of atherosclerosis mechanisms, and has the potential to be a novel index of arterial stiffness.
Address
Corporate Author Thesis
Publisher (up) Place of Publication Editor
Language Summary Language Original Title
Series Editor Series Title Abbreviated Series Title
Series Volume Series Issue Edition
ISSN ISBN Medium
Area Expedition Conference
Notes MILAB Approved no
Call Number Admin @ si @ ZBS2014 Serial 2556
Permanent link to this record
 
 
Author G. Lisanti; I. Masi; Andrew Bagdanov; Alberto del Bimbo
Title Person Re-identification by Iterative Re-weighted Sparse Ranking Type Journal Article
Year 2015 Publication IEEE Transactions on Pattern Analysis and Machine Intelligence Abbreviated Journal TPAMI
Volume 37 Issue 8 Pages 1629 - 1642
Keywords
Abstract In this paper we introduce a method for person re-identification based on discriminative, sparse basis expansions of targets in terms of a labeled gallery of known individuals. We propose an iterative extension to sparse discriminative classifiers capable of ranking many candidate targets. The approach makes use of soft- and hard- re-weighting to redistribute energy among the most relevant contributing elements and to ensure that the best candidates are ranked at each iteration. Our approach also leverages a novel visual descriptor which we show to be discriminative while remaining robust to pose and illumination variations. An extensive comparative evaluation is given demonstrating that our approach achieves state-of-the-art performance on single- and multi-shot person re-identification scenarios on the VIPeR, i-LIDS, ETHZ, and CAVIAR4REID datasets. The combination of our descriptor and iterative sparse basis expansion improves state-of-the-art rank-1 performance by six percentage points on VIPeR and by 20 on CAVIAR4REID compared to other methods with a single gallery image per person. With multiple gallery and probe images per person our approach improves by 17 percentage points the state-of-the-art on i-LIDS and by 72 on CAVIAR4REID at rank-1. The approach is also quite efficient, capable of single-shot person re-identification over galleries containing hundreds of individuals at about 30 re-identifications per second.
Address
Corporate Author Thesis
Publisher (up) Place of Publication Editor
Language Summary Language Original Title
Series Editor Series Title Abbreviated Series Title
Series Volume Series Issue Edition
ISSN 0162-8828 ISBN Medium
Area Expedition Conference
Notes LAMP; 601.240; 600.079 Approved no
Call Number Admin @ si @ LMB2015 Serial 2557
Permanent link to this record
 
 
Author Alejandro Gonzalez Alzate; Gabriel Villalonga; Jiaolong Xu; David Vazquez; Jaume Amores; Antonio Lopez
Title Multiview Random Forest of Local Experts Combining RGB and LIDAR data for Pedestrian Detection Type Conference Article
Year 2015 Publication IEEE Intelligent Vehicles Symposium IV2015 Abbreviated Journal
Volume Issue Pages 356-361
Keywords Pedestrian Detection
Abstract Despite recent significant advances, pedestrian detection continues to be an extremely challenging problem in real scenarios. In order to develop a detector that successfully operates under these conditions, it becomes critical to leverage upon multiple cues, multiple imaging modalities and a strong multi-view classifier that accounts for different pedestrian views and poses. In this paper we provide an extensive evaluation that gives insight into how each of these aspects (multi-cue, multimodality and strong multi-view classifier) affect performance both individually and when integrated together. In the multimodality component we explore the fusion of RGB and depth maps obtained by high-definition LIDAR, a type of modality that is only recently starting to receive attention. As our analysis reveals, although all the aforementioned aspects significantly help in improving the performance, the fusion of visible spectrum and depth information allows to boost the accuracy by a much larger margin. The resulting detector not only ranks among the top best performers in the challenging KITTI benchmark, but it is built upon very simple blocks that are easy to implement and computationally efficient. These simple blocks can be easily replaced with more sophisticated ones recently proposed, such as the use of convolutional neural networks for feature representation, to further improve the accuracy.
Address Seoul; Corea; June 2015
Corporate Author Thesis
Publisher (up) Place of Publication Editor
Language Summary Language Original Title
Series Editor Series Title Abbreviated Series Title
Series Volume Series Issue Edition
ISSN ISBN Medium
Area ACDC Expedition Conference IV
Notes ADAS; 600.076; 600.057; 600.054 Approved no
Call Number ADAS @ adas @ GVX2015 Serial 2625
Permanent link to this record
 
 
Author Mikhail Mozerov; Joost Van de Weijer
Title Accurate stereo matching by two step global optimization Type Journal Article
Year 2015 Publication IEEE Transactions on Image Processing Abbreviated Journal TIP
Volume 24 Issue 3 Pages 1153-1163
Keywords
Abstract In stereo matching cost filtering methods and energy minimization algorithms are considered as two different techniques. Due to their global extend energy minimization methods obtain good stereo matching results. However, they tend to fail in occluded regions, in which cost filtering approaches obtain better results. In this paper we intend to combine both approaches with the aim to improve overall stereo matching results. We show that a global optimization with a fully connected model can be solved by cost fil tering methods. Based on this observation we propose to perform stereo matching as a two-step energy minimization algorithm. We consider two MRF models: a fully connected model defined on the complete set of pixels in an image and a conventional locally connected model. We solve the energy minimization problem for the fully connected model, after which the marginal function of the solution is used as the unary potential in the locally connected MRF model. Experiments on the Middlebury stereo datasets show that the proposed method achieves state-of-the-arts results.
Address
Corporate Author Thesis
Publisher (up) Place of Publication Editor
Language Summary Language Original Title
Series Editor Series Title Abbreviated Series Title
Series Volume Series Issue Edition
ISSN 1057-7149 ISBN Medium
Area Expedition Conference
Notes ISE; LAMP; 600.079; 600.078 Approved no
Call Number Admin @ si @ MoW2015a Serial 2568
Permanent link to this record
 
 
Author G.Thorvaldsen; Joana Maria Pujadas-Mora; T.Andersen ; L.Eikvil; Josep Llados; Alicia Fornes; Anna Cabre
Title A Tale of two Transcriptions Type Journal
Year 2015 Publication Historical Life Course Studies Abbreviated Journal
Volume 2 Issue Pages 1-19
Keywords Nominative Sources; Census; Vital Records; Computer Vision; Optical Character Recognition; Word Spotting
Abstract non-indexed
This article explains how two projects implement semi-automated transcription routines: for census sheets in Norway and marriage protocols from Barcelona. The Spanish system was created to transcribe the marriage license books from 1451 to 1905 for the Barcelona area; one of the world’s longest series of preserved vital records. Thus, in the Project “Five Centuries of Marriages” (5CofM) at the Autonomous University of Barcelona’s Center for Demographic Studies, the Barcelona Historical Marriage Database has been built. More than 600,000 records were transcribed by 150 transcribers working online. The Norwegian material is cross-sectional as it is the 1891 census, recorded on one sheet per person. This format and the underlining of keywords for several variables made it more feasible to semi-automate data entry than when many persons are listed on the same page. While Optical Character Recognition (OCR) for printed text is scientifically mature, computer vision research is now focused on more difficult problems such as handwriting recognition. In the marriage project, document analysis methods have been proposed to automatically recognize the marriage licenses. Fully automatic recognition is still a challenge, but some promising results have been obtained. In Spain, Norway and elsewhere the source material is available as scanned pictures on the Internet, opening up the possibility for further international cooperation concerning automating the transcription of historic source materials. Like what is being done in projects to digitize printed materials, the optimal solution is likely to be a combination of manual transcription and machine-assisted recognition also for hand-written sources.
Address
Corporate Author Thesis
Publisher (up) Place of Publication Editor
Language Summary Language Original Title
Series Editor Series Title Abbreviated Series Title
Series Volume Series Issue Edition
ISSN 2352-6343 ISBN Medium
Area Expedition Conference
Notes DAG; 600.077; 602.006 Approved no
Call Number Admin @ si @ TPA2015 Serial 2582
Permanent link to this record
 
 
Author Alejandro Gonzalez Alzate; Gabriel Villalonga; German Ros; David Vazquez; Antonio Lopez
Title 3D-Guided Multiscale Sliding Window for Pedestrian Detection Type Conference Article
Year 2015 Publication Pattern Recognition and Image Analysis, Proceedings of 7th Iberian Conference , ibPRIA 2015 Abbreviated Journal
Volume 9117 Issue Pages 560-568
Keywords Pedestrian Detection
Abstract The most relevant modules of a pedestrian detector are the candidate generation and the candidate classification. The former aims at presenting image windows to the latter so that they are classified as containing a pedestrian or not. Much attention has being paid to the classification module, while candidate generation has mainly relied on (multiscale) sliding window pyramid. However, candidate generation is critical for achieving real-time. In this paper we assume a context of autonomous driving based on stereo vision. Accordingly, we evaluate the effect of taking into account the 3D information (derived from the stereo) in order to prune the hundred of thousands windows per image generated by classical pyramidal sliding window. For our study we use a multimodal (RGB, disparity) and multi-descriptor (HOG, LBP, HOG+LBP) holistic ensemble based on linear SVM. Evaluation on data from the challenging KITTI benchmark suite shows the effectiveness of using 3D information to dramatically reduce the number of candidate windows, even improving the overall pedestrian detection accuracy.
Address Santiago de Compostela; España; June 2015
Corporate Author Thesis
Publisher (up) Place of Publication Editor
Language Summary Language Original Title
Series Editor Series Title Abbreviated Series Title
Series Volume Series Issue Edition
ISSN ISBN Medium
Area ACDC Expedition Conference IbPRIA
Notes ADAS; 600.076; 600.057; 600.054 Approved no
Call Number ADAS @ adas @ GVR2015 Serial 2585
Permanent link to this record
 
 
Author Joost Van de Weijer; Fahad Shahbaz Khan
Title An Overview of Color Name Applications in Computer Vision Type Conference Article
Year 2015 Publication Computational Color Imaging Workshop Abbreviated Journal
Volume Issue Pages
Keywords color features; color names; object recognition
Abstract In this article we provide an overview of color name applications in computer vision. Color names are linguistic labels which humans use to communicate color. Computational color naming learns a mapping from pixels values to color names. In recent years color names have been applied to a wide variety of computer vision applications, including image classification, object recognition, texture classification, visual tracking and action recognition. Here we provide an overview of these results which show that in general color names outperform photometric invariants as a color representation.
Address Saint Etienne; France; March 2015
Corporate Author Thesis
Publisher (up) Place of Publication Editor
Language Summary Language Original Title
Series Editor Series Title Abbreviated Series Title
Series Volume Series Issue Edition
ISSN ISBN Medium
Area Expedition Conference CCIW
Notes LAMP; 600.079; 600.068 Approved no
Call Number Admin @ si @ WeK2015 Serial 2586
Permanent link to this record
 
 
Author Fahad Shahbaz Khan; Muhammad Anwer Rao; Joost Van de Weijer; Michael Felsberg; J.Laaksonen
Title Compact color texture description for texture classification Type Journal Article
Year 2015 Publication Pattern Recognition Letters Abbreviated Journal PRL
Volume 51 Issue Pages 16-22
Keywords
Abstract Describing textures is a challenging problem in computer vision and pattern recognition. The classification problem involves assigning a category label to the texture class it belongs to. Several factors such as variations in scale, illumination and viewpoint make the problem of texture description extremely challenging. A variety of histogram based texture representations exists in literature.
However, combining multiple texture descriptors and assessing their complementarity is still an open research problem. In this paper, we first show that combining multiple local texture descriptors significantly improves the recognition performance compared to using a single best method alone. This
gain in performance is achieved at the cost of high-dimensional final image representation. To counter this problem, we propose to use an information-theoretic compression technique to obtain a compact texture description without any significant loss in accuracy. In addition, we perform a comprehensive
evaluation of pure color descriptors, popular in object recognition, for the problem of texture classification. Experiments are performed on four challenging texture datasets namely, KTH-TIPS-2a, KTH-TIPS-2b, FMD and Texture-10. The experiments clearly demonstrate that our proposed compact multi-texture approach outperforms the single best texture method alone. In all cases, discriminative color names outperforms other color features for texture classification. Finally, we show that combining discriminative color names with compact texture representation outperforms state-of-the-art methods by 7:8%, 4:3% and 5:0% on KTH-TIPS-2a, KTH-TIPS-2b and Texture-10 datasets respectively.
Address
Corporate Author Thesis
Publisher (up) Place of Publication Editor
Language Summary Language Original Title
Series Editor Series Title Abbreviated Series Title
Series Volume Series Issue Edition
ISSN ISBN Medium
Area Expedition Conference
Notes LAMP; 600.068; 600.079;ADAS Approved no
Call Number Admin @ si @ KRW2015a Serial 2587
Permanent link to this record
 
 
Author Meysam Madadi; Sergio Escalera; Jordi Gonzalez; Xavier Roca; Felipe Lumbreras
Title Multi-part body segmentation based on depth maps for soft biometry analysis Type Journal Article
Year 2015 Publication Pattern Recognition Letters Abbreviated Journal PRL
Volume 56 Issue Pages 14-21
Keywords 3D shape context; 3D point cloud alignment; Depth maps; Human body segmentation; Soft biometry analysis
Abstract This paper presents a novel method extracting biometric measures using depth sensors. Given a multi-part labeled training data, a new subject is aligned to the best model of the dataset, and soft biometrics such as lengths or circumference sizes of limbs and body are computed. The process is performed by training relevant pose clusters, defining a representative model, and fitting a 3D shape context descriptor within an iterative matching procedure. We show robust measures by applying orthogonal plates to body hull. We test our approach in a novel full-body RGB-Depth data set, showing accurate estimation of soft biometrics and better segmentation accuracy in comparison with random forest approach without requiring large training data.
Address
Corporate Author Thesis
Publisher (up) Place of Publication Editor
Language Summary Language Original Title
Series Editor Series Title Abbreviated Series Title
Series Volume Series Issue Edition
ISSN ISBN Medium
Area Expedition Conference
Notes HuPBA; ISE; ADAS; 600.076;600.049; 600.063; 600.054; 302.018;MILAB Approved no
Call Number Admin @ si @ MEG2015 Serial 2588
Permanent link to this record
 
 
Author Ivan Huerta; Marco Pedersoli; Jordi Gonzalez; Alberto Sanfeliu
Title Combining where and what in change detection for unsupervised foreground learning in surveillance Type Journal Article
Year 2015 Publication Pattern Recognition Abbreviated Journal PR
Volume 48 Issue 3 Pages 709-719
Keywords Object detection; Unsupervised learning; Motion segmentation; Latent variables; Support vector machine; Multiple appearance models; Video surveillance
Abstract Change detection is the most important task for video surveillance analytics such as foreground and anomaly detection. Current foreground detectors learn models from annotated images since the goal is to generate a robust foreground model able to detect changes in all possible scenarios. Unfortunately, manual labelling is very expensive. Most advanced supervised learning techniques based on generic object detection datasets currently exhibit very poor performance when applied to surveillance datasets because of the unconstrained nature of such environments in terms of types and appearances of objects. In this paper, we take advantage of change detection for training multiple foreground detectors in an unsupervised manner. We use statistical learning techniques which exploit the use of latent parameters for selecting the best foreground model parameters for a given scenario. In essence, the main novelty of our proposed approach is to combine the where (motion segmentation) and what (learning procedure) in change detection in an unsupervised way for improving the specificity and generalization power of foreground detectors at the same time. We propose a framework based on latent support vector machines that, given a noisy initialization based on motion cues, learns the correct position, aspect ratio, and appearance of all moving objects in a particular scene. Specificity is achieved by learning the particular change detections of a given scenario, and generalization is guaranteed since our method can be applied to any possible scene and foreground object, as demonstrated in the experimental results outperforming the state-of-the-art.
Address
Corporate Author Thesis
Publisher (up) Place of Publication Editor
Language Summary Language Original Title
Series Editor Series Title Abbreviated Series Title
Series Volume Series Issue Edition
ISSN ISBN Medium
Area Expedition Conference
Notes ISE; 600.063; 600.078 Approved no
Call Number Admin @ si @ HPG2015 Serial 2589
Permanent link to this record
 
 
Author Wenjuan Gong; Y.Huang; Jordi Gonzalez; Liang Wang
Title An Effective Solution to Double Counting Problem in Human Pose Estimation Type Miscellaneous
Year 2015 Publication Arxiv Abbreviated Journal
Volume Issue Pages
Keywords Pose estimation; double counting problem; mix-ture of parts Model
Abstract The mixture of parts model has been successfully applied to solve the 2D
human pose estimation problem either as an explicitly trained body part model
or as latent variables for pedestrian detection. Even in the era of massive
applications of deep learning techniques, the mixture of parts model is still
effective in solving certain problems, especially in the case with limited
numbers of training samples. In this paper, we consider using the mixture of
parts model for pose estimation, wherein a tree structure is utilized for
representing relations between connected body parts. This strategy facilitates
training and inferencing of the model but suffers from double counting
problems, where one detected body part is counted twice due to lack of
constrains among unconnected body parts. To solve this problem, we propose a
generalized solution in which various part attributes are captured by multiple
features so as to avoid the double counted problem. Qualitative and
quantitative experimental results on a public available dataset demonstrate the
effectiveness of our proposed method.

An Effective Solution to Double Counting Problem in Human Pose Estimation – ResearchGate. Available from: http://www.researchgate.net/publication/271218491AnEffectiveSolutiontoDoubleCountingProbleminHumanPose_Estimation [accessed Oct 22, 2015].
Address
Corporate Author Thesis
Publisher (up) Place of Publication Editor
Language Summary Language Original Title
Series Editor Series Title Abbreviated Series Title
Series Volume Series Issue Edition
ISSN ISBN Medium
Area Expedition Conference
Notes ISE; 600.078 Approved no
Call Number Admin @ si @ GHG2015 Serial 2590
Permanent link to this record
 
 
Author Sergio Escalera; Jordi Gonzalez; Xavier Baro; Pablo Pardo; Junior Fabian; Marc Oliu; Hugo Jair Escalante; Ivan Huerta; Isabelle Guyon
Title ChaLearn Looking at People 2015 new competitions: Age Estimation and Cultural Event Recognition Type Conference Article
Year 2015 Publication IEEE International Joint Conference on Neural Networks IJCNN2015 Abbreviated Journal
Volume Issue Pages 1-8
Keywords
Abstract Following previous series on Looking at People (LAP) challenges [1], [2], [3], in 2015 ChaLearn runs two new competitions within the field of Looking at People: age and cultural event recognition in still images. We propose thefirst crowdsourcing application to collect and label data about apparent
age of people instead of the real age. In terms of cultural event recognition, tens of categories have to be recognized. This involves scene understanding and human analysis. This paper summarizes both challenges and data, providing some initial baselines. The results of the first round of the competition were presented at ChaLearn LAP 2015 IJCNN special session on computer vision and robotics http://www.dtic.ua.es/∼jgarcia/IJCNN2015.
Details of the ChaLearn LAP competitions can be found at http://gesture.chalearn.org/.
Address Killarney; Ireland; July 2015
Corporate Author Thesis
Publisher (up) Place of Publication Editor
Language Summary Language Original Title
Series Editor Series Title Abbreviated Series Title
Series Volume Series Issue Edition
ISSN ISBN Medium
Area Expedition Conference IJCNN
Notes HuPBA; ISE; 600.063; 600.078;MV Approved no
Call Number Admin @ si @ EGB2015 Serial 2591
Permanent link to this record
 
 
Author Frederic Sampedro; Anna Domenech; Sergio Escalera; Ignasi Carrio
Title Deriving global quantitative tumor response parameters from 18F-FDG PET-CT scans in patients with non-Hodgkins lymphoma Type Journal Article
Year 2015 Publication Nuclear Medicine Communications Abbreviated Journal NMC
Volume 36 Issue 4 Pages 328-333
Keywords
Abstract OBJECTIVES:
The aim of the study was to address the need for quantifying the global cancer time evolution magnitude from a pair of time-consecutive positron emission tomography-computed tomography (PET-CT) scans. In particular, we focus on the computation of indicators using image-processing techniques that seek to model non-Hodgkin's lymphoma (NHL) progression or response severity.
MATERIALS AND METHODS:
A total of 89 pairs of time-consecutive PET-CT scans from NHL patients were stored in a nuclear medicine station for subsequent analysis. These were classified by a consensus of nuclear medicine physicians into progressions, partial responses, mixed responses, complete responses, and relapses. The cases of each group were ordered by magnitude following visual analysis. Thereafter, a set of quantitative indicators designed to model the cancer evolution magnitude within each group were computed using semiautomatic and automatic image-processing techniques. Performance evaluation of the proposed indicators was measured by a correlation analysis with the expert-based visual analysis.
RESULTS:
The set of proposed indicators achieved Pearson's correlation results in each group with respect to the expert-based visual analysis: 80.2% in progressions, 77.1% in partial response, 68.3% in mixed response, 88.5% in complete response, and 100% in relapse. In the progression and mixed response groups, the proposed indicators outperformed the common indicators used in clinical practice [changes in metabolic tumor volume, mean, maximum, peak standardized uptake value (SUV mean, SUV max, SUV peak), and total lesion glycolysis] by more than 40%.
CONCLUSION:
Computing global indicators of NHL response using PET-CT imaging techniques offers a strong correlation with the associated expert-based visual analysis, motivating the future incorporation of such quantitative and highly observer-independent indicators in oncological decision making or treatment response evaluation scenarios.
Address
Corporate Author Thesis
Publisher (up) Place of Publication Editor
Language Summary Language Original Title
Series Editor Series Title Abbreviated Series Title
Series Volume Series Issue Edition
ISSN ISBN Medium
Area Expedition Conference
Notes HuPBA;MILAB Approved no
Call Number Admin @ si @ SDE2015 Serial 2605
Permanent link to this record
 
 
Author Wenjuan Gong; W.Zhang; Jordi Gonzalez; Y.Ren; Z.Li
Title Enhanced Asymmetric Bilinear Model for Face Recognition Type Journal Article
Year 2015 Publication International Journal of Distributed Sensor Networks Abbreviated Journal IJDSN
Volume Issue Pages Article ID 218514
Keywords
Abstract Bilinear models have been successfully applied to separate two factors, for example, pose variances and different identities in face recognition problems. Asymmetric model is a type of bilinear model which models a system in the most concise way. But seldom there are works exploring the applications of asymmetric bilinear model on face recognition problem with illumination changes. In this work, we propose enhanced asymmetric model for illumination-robust face recognition. Instead of initializing the factor probabilities randomly, we initialize them with nearest neighbor method and optimize them for the test data. Above that, we update the factor model to be identified. We validate the proposed method on a designed data sample and extended Yale B dataset. The experiment results show that the enhanced asymmetric models give promising results and good recognition accuracies.
Address
Corporate Author Thesis
Publisher (up) Place of Publication Editor
Language Summary Language Original Title
Series Editor Series Title Abbreviated Series Title
Series Volume Series Issue Edition
ISSN ISBN Medium
Area Expedition Conference
Notes ISE; 600.063; 600.078 Approved no
Call Number Admin @ si @ GZG2015 Serial 2592
Permanent link to this record
 
 
Author Adriana Romero; Nicolas Ballas; Samira Ebrahimi Kahou; Antoine Chassang; Carlo Gatta; Yoshua Bengio
Title FitNets: Hints for Thin Deep Nets Type Conference Article
Year 2015 Publication 3rd International Conference on Learning Representations ICLR2015 Abbreviated Journal
Volume Issue Pages
Keywords Computer Science ; Learning; Computer Science ;Neural and Evolutionary Computing
Abstract While depth tends to improve network performances, it also makes gradient-based training more difficult since deeper networks tend to be more non-linear. The recently proposed knowledge distillation approach is aimed at obtaining small and fast-to-execute models, and it has shown that a student network could imitate the soft output of a larger teacher network or ensemble of networks. In this paper, we extend this idea to allow the training of a student that is deeper and thinner than the teacher, using not only the outputs but also the intermediate representations learned by the teacher as hints to improve the training process and final performance of the student. Because the student intermediate hidden layer will generally be smaller than the teacher's intermediate hidden layer, additional parameters are introduced to map the student hidden layer to the prediction of the teacher hidden layer. This allows one to train deeper students that can generalize better or run faster, a trade-off that is controlled by the chosen student capacity. For example, on CIFAR-10, a deep student network with almost 10.4 times less parameters outperforms a larger, state-of-the-art teacher network.
Address San Diego; CA; May 2015
Corporate Author Thesis
Publisher (up) Place of Publication Editor
Language Summary Language Original Title
Series Editor Series Title Abbreviated Series Title
Series Volume Series Issue Edition
ISSN ISBN Medium
Area Expedition Conference ICLR
Notes MILAB Approved no
Call Number Admin @ si @ RBK2015 Serial 2593
Permanent link to this record