{"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/learning-to-learn-with-backpropagation-of","title":"Learning to learn with backpropagation of Hebbian plasticity","arxiv_id":"1609.02228","date":"2016-09-08","proceeding":null,"authors":["Thomas Miconi"],"abstract":"Hebbian plasticity is a powerful principle that allows biological brains to\nlearn from their lifetime experience. By contrast, artificial neural networks\ntrained with backpropagation generally have fixed connection weights that do\nnot change once training is complete. While recent methods can endow neural\nnetworks with long-term memories, Hebbian plasticity is currently not amenable\nto gradient descent. Here we derive analytical expressions for activity\ngradients in neural networks with Hebbian plastic connections. Using these\nexpressions, we can use backpropagation to train not just the baseline weights\nof the connections, but also their plasticity. As a result, the networks \"learn\nhow to learn\" in order to solve the problem at hand: the trained networks\nautomatically perform fast learning of unpredictable environmental features\nduring their lifetime, expanding the range of solvable problems. We test the\nalgorithm on various on-line learning tasks, including pattern completion,\none-shot learning, and reversal learning. The algorithm successfully learns how\nto learn the relevant associations from one-shot instruction, and fine-tunes\nthe temporal dynamics of plasticity to allow for continual learning in response\nto changing environmental parameters. We conclude that backpropagation of\nHebbian plasticity offers a powerful model for lifelong learning.","url_abs":"http://arxiv.org/abs/1609.02228v2","url_pdf":"http://arxiv.org/pdf/1609.02228v2.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":"learning-to-learn-with-backpropagation-of","repo_url":"https://github.com/ThomasMiconi/LearningToLearnBOHP","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"continual-learning","task_name":"Continual Learning"},{"task_slug":"lifelong-learning","task_name":"Lifelong learning"},{"task_slug":"one-shot-learning","task_name":"One-Shot Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}