TY - JOUR
T1 - Curriculum learning of visual attribute clusters for multi-task classification
AU - Sarafianos, Nikolaos
AU - Giannakopoulos, Theodoros
AU - Nikou, Christophoros
AU - Kakadiaris, Ioannis A.
N1 - Funding Information:
Ioannis A. Kakadiaris is a Hugh Roy and Lillie Cranz Cullen University Professor of Computer Science, Electrical & Computer Engineering, and Biomedical Engineering at the University of Houston. He joined UH in 1997 after a postdoctoral fellowship at the University of Pennsylvania. He earned his B.Sc. in physics at the University of Athens in Greece, his M.Sc. in computer science from Northeastern University and his Ph.D. at the University of Pennsylvania. He is the founder of the Computational Biomedicine Lab and the Director of the DHS Center of Excellence on Borders, Trade, and Immigration Research (BTI). His research interests include biometrics, video analytics, computer vision, pattern recognition, and biomedical image computing. He is the recipient of a number of awards, including the NSF Early Career Development Award, Schlumberger Technical Foundation Award, UH Computer Science Research Excellence Award, UH Enron Teaching Excellence Award, and the James Muller Vulnerable Plaque Young Investigator Prize. His research has been featured on The Discovery Channel, National Public Radio, KPRC NBC News, KTRH ABC News, and KHOU CBS News.
Funding Information:
This work has been funded in part by the UH Hugh Roy and Lillie Cranz Cullen Endowment Fund. The work of C. Nikou is supported by the European Commission (H2020-MSCA-IF-2014), under grant agreement no 656094 . All statements of fact, opinion or conclusions contained herein are those of the authors and should not be construed as representing the official views or policies of the sponsors.
Publisher Copyright:
© 2018 Elsevier Ltd
PY - 2018/8
Y1 - 2018/8
N2 - Visual attributes, from simple objects (e.g., backpacks, hats) to soft-biometrics (e.g., gender, height, clothing) have proven to be a powerful representational approach for many applications such as image description and human identification. In this paper, we introduce a novel method to combine the advantages of both multi-task and curriculum learning in a visual attribute classification framework. Individual tasks are grouped after performing hierarchical clustering based on their correlation. The clusters of tasks are learned in a curriculum learning setup by transferring knowledge between clusters. The learning process within each cluster is performed in a multi-task classification setup. By leveraging the acquired knowledge, we speed-up the process and improve performance. We demonstrate the effectiveness of our method via ablation studies and a detailed analysis of the covariates, on a variety of publicly available datasets of humans standing with their full-body visible. Extensive experimentation has proven that the proposed approach boosts the performance by 4%–10%.
AB - Visual attributes, from simple objects (e.g., backpacks, hats) to soft-biometrics (e.g., gender, height, clothing) have proven to be a powerful representational approach for many applications such as image description and human identification. In this paper, we introduce a novel method to combine the advantages of both multi-task and curriculum learning in a visual attribute classification framework. Individual tasks are grouped after performing hierarchical clustering based on their correlation. The clusters of tasks are learned in a curriculum learning setup by transferring knowledge between clusters. The learning process within each cluster is performed in a multi-task classification setup. By leveraging the acquired knowledge, we speed-up the process and improve performance. We demonstrate the effectiveness of our method via ablation studies and a detailed analysis of the covariates, on a variety of publicly available datasets of humans standing with their full-body visible. Extensive experimentation has proven that the proposed approach boosts the performance by 4%–10%.
KW - Curriculum learning
KW - Multi-task classification
KW - Visual attributes
UR - http://www.scopus.com/inward/record.url?scp=85046036875&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85046036875&partnerID=8YFLogxK
U2 - 10.1016/j.patcog.2018.02.028
DO - 10.1016/j.patcog.2018.02.028
M3 - Article
AN - SCOPUS:85046036875
VL - 80
SP - 94
EP - 108
JO - Pattern Recognition
JF - Pattern Recognition
SN - 0031-3203
ER -