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

Considering the instance-level discriminative ability, contrastive learningmethods, including MoCo and SimCLR, have been adapted from the original imagerepresentation learning task to solve the self-supervised skeleton-based actionrecognition task. These methods usually use multiple data streams (i.e., joint,motion, and bone) for ensemble learning, meanwhile, how to construct adiscriminative feature space within a single stream and effectively aggregatethe information from multiple streams remains an open problem. To this end, wefirst apply a new contrastive learning method called BYOL to learn fromskeleton data and formulate SkeletonBYOL as a simple yet effective baseline forself-supervised skeleton-based action recognition. Inspired by SkeletonBYOL, wefurther present a joint Adversarial and Collaborative Learning (ACL) framework,which combines Cross-Model Adversarial Learning (CMAL) and Cross-StreamCollaborative Learning (CSCL). Specifically, CMAL learns single-streamrepresentation by cross-model adversarial loss to obtain more discriminativefeatures. To aggregate and interact with multi-stream information, CSCL isdesigned by generating similarity pseudo label of ensemble learning assupervision and guiding feature generation for individual streams. Exhaustiveexperiments on three datasets verify the complementary properties between CMALand CSCL and also verify that our method can perform favorably againststate-of-the-art methods using various evaluation protocols. Our code andmodels are publicly available at \url{https://github.com/Levigty/ACL}.

Joint Adversarial and Collaborative Learning for Self-Supervised Action Recognition