English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Video representation learning is a vital problem for classification task.Recently, a promising unsupervised paradigm termed self-supervised learning hasemerged, which explores inherent supervisory signals implied in massive datafor feature learning via solving auxiliary tasks. However, existing methods inthis regard suffer from two limitations when extended to video classification.First, they focus only on a single task, whereas ignoring complementarity amongdifferent task-specific features and thus resulting in suboptimal videorepresentation. Second, high computational and memory cost hinders theirapplication in real-world scenarios. In this paper, we propose a graph-baseddistillation framework to address these problems: (1) We propose logits graphand representation graph to transfer knowledge from multiple self-supervisedtasks, where the former distills classifier-level knowledge by solving amulti-distribution joint matching problem, and the latter distills internalfeature knowledge from pairwise ensembled representations with tackling thechallenge of heterogeneity among different features; (2) The proposal thatadopts a teacher-student framework can reduce the redundancy of knowledgelearnt from teachers dramatically, leading to a lighter student model thatsolves classification task more efficiently. Experimental results on 3 videodatasets validate that our proposal not only helps learn better videorepresentation but also compress model for faster inference.

Better and Faster: Knowledge Transfer from Multiple Self-supervised Learning Tasks via Graph Distillation for Video Classification