Curriculum learning of visual attribute clusters for multi-task classification. (August 2018)

• Record Type: Journal Article
• Authors: Sarafianos, Nikolaos; Giannakopoulos, Theodoros; Nikou, Christophoros; Kakadiaris, Ioannis A.
• Is Part Of: Pattern recognition. Volume 80(2018:Aug.)
• Page Start: 94
• Page End: 108
• Publication Date: 2018-08
• Subjects: Curriculum learning -- Multi-task classification -- Visual attributes; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2018.02.028
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: Novel learning scheme that combines the advantages of multi-task and curriculum learning. Efficient correlation-based grouping of tasks based on hierarchical clustering. Extensive evaluation, analysis of covariates and ablation studies. Abstract: 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%.