{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/equivalence-of-equilibrium-propagation-and","title":"Equivalence of Equilibrium Propagation and Recurrent Backpropagation","arxiv_id":"1711.08416","date":"2017-11-22","proceeding":null,"authors":["Benjamin Scellier","Yoshua Bengio"],"abstract":"Recurrent Backpropagation and Equilibrium Propagation are supervised learning\nalgorithms for fixed point recurrent neural networks which differ in their\nsecond phase. In the first phase, both algorithms converge to a fixed point\nwhich corresponds to the configuration where the prediction is made. In the\nsecond phase, Equilibrium Propagation relaxes to another nearby fixed point\ncorresponding to smaller prediction error, whereas Recurrent Backpropagation\nuses a side network to compute error derivatives iteratively. In this work we\nestablish a close connection between these two algorithms. We show that, at\nevery moment in the second phase, the temporal derivatives of the neural\nactivities in Equilibrium Propagation are equal to the error derivatives\ncomputed iteratively by Recurrent Backpropagation in the side network. This\nwork shows that it is not required to have a side network for the computation\nof error derivatives, and supports the hypothesis that, in biological neural\nnetworks, temporal derivatives of neural activities may code for error signals.","url_abs":"http://arxiv.org/abs/1711.08416v2","url_pdf":"http://arxiv.org/pdf/1711.08416v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"equivalence-of-equilibrium-propagation-and","repo_url":"https://github.com/bscellier/Towards-a-Biologically-Plausible-Backprop","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1711.08416","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}