{"id":"arxiv:2103.11275","kind":"preprint","title":"Self-supervised Representation Learning with Relative Predictive Coding","authors":["Yao-Hung Hubert Tsai","Martin Q. Ma","Muqiao Yang","Han Zhao","Louis-Philippe Morency","Ruslan Salakhutdinov"],"author_count":6,"year":2021,"published":"2021-03-21","venue":null,"cited_by":9,"doi":null,"arxiv_id":"2103.11275","nct_id":null,"source_url":"https://arxiv.org/abs/2103.11275","archive_url":"https://space-kitty.com/papers/arxiv:2103.11275","updated":"2021-04-12","pmid":null,"pmcid":null,"openalex_id":"W3124635886","pdf_url":"https://arxiv.org/pdf/2103.11275","abstract":"This paper introduces Relative Predictive Coding (RPC), a new contrastive representation learning objective that maintains a good balance among training stability, minibatch size sensitivity, and downstream task performance. The key to the success of RPC is two-fold. First, RPC introduces the relative parameters to regularize the objective for boundedness and low variance. Second, RPC contains no logarithm and exponential score functions, which are the main cause of training instability in prior contrastive objectives. We empirically verify the effectiveness of RPC on benchmark vision and speech self-supervised learning tasks. Lastly, we relate RPC with mutual information (MI) estimation, showing RPC can be used to estimate MI with low variance.","abstract_license":"CC0-1.0 (arXiv descriptive metadata)","license":null,"is_open_access":false,"retracted":false,"peer_reviewed":false,"categories":["cs.LG","cs.IT"],"details":null,"provenance":"arxiv","attribution":"Abstract: arXiv descriptive metadata (CC0). Thank you to arXiv for use of its open access interoperability. This service was not reviewed or approved by, nor does it necessarily express or reflect the policies or opinions of, arXiv.","topics":[{"slug":"neuroai","name":"NeuroAI & Computational Neuroscience","domain_slug":"neuroscience"}],"cited_in_dossiers":[]}