{"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/conditional-computation-in-neural-networks","title":"Conditional Computation in Neural Networks for faster models","arxiv_id":"1511.06297","date":"2015-11-19","proceeding":null,"authors":["Emmanuel Bengio","Pierre-Luc Bacon","Joelle Pineau","Doina Precup"],"abstract":"Deep learning has become the state-of-art tool in many applications, but the\nevaluation and training of deep models can be time-consuming and\ncomputationally expensive. The conditional computation approach has been\nproposed to tackle this problem (Bengio et al., 2013; Davis & Arel, 2013). It\noperates by selectively activating only parts of the network at a time. In this\npaper, we use reinforcement learning as a tool to optimize conditional\ncomputation policies. More specifically, we cast the problem of learning\nactivation-dependent policies for dropping out blocks of units as a\nreinforcement learning problem. We propose a learning scheme motivated by\ncomputation speed, capturing the idea of wanting to have parsimonious\nactivations while maintaining prediction accuracy. We apply a policy gradient\nalgorithm for learning policies that optimize this loss function and propose a\nregularization mechanism that encourages diversification of the dropout policy.\nWe present encouraging empirical results showing that this approach improves\nthe speed of computation without impacting the quality of the approximation.","url_abs":"http://arxiv.org/abs/1511.06297v2","url_pdf":"http://arxiv.org/pdf/1511.06297v2.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":"conditional-computation-in-neural-networks","repo_url":"https://github.com/bengioe/condnet","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"},{"task_slug":"reinforcement-learning-2","task_name":"reinforcement-learning"}],"methods":[{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"speed","method_name":"SPEED"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1511.06297","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}