{"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-continuous-control-policies-by","title":"Learning Continuous Control Policies by Stochastic Value Gradients","arxiv_id":"1510.09142","date":"2015-10-30","proceeding":"NeurIPS 2015 12","authors":["Nicolas Heess","Greg Wayne","David Silver","Timothy Lillicrap","Yuval Tassa","Tom Erez"],"abstract":"We present a unified framework for learning continuous control policies using\nbackpropagation. It supports stochastic control by treating stochasticity in\nthe Bellman equation as a deterministic function of exogenous noise. The\nproduct is a spectrum of general policy gradient algorithms that range from\nmodel-free methods with value functions to model-based methods without value\nfunctions. We use learned models but only require observations from the\nenvironment in- stead of observations from model-predicted trajectories,\nminimizing the impact of compounded model errors. We apply these algorithms\nfirst to a toy stochastic control problem and then to several physics-based\ncontrol problems in simulation. One of these variants, SVG(1), shows the\neffectiveness of learning models, value functions, and policies simultaneously\nin continuous domains.","url_abs":"http://arxiv.org/abs/1510.09142v1","url_pdf":"http://arxiv.org/pdf/1510.09142v1.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-continuous-control-policies-by","repo_url":"https://github.com/RobvanGastel/svg-priors","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"learning-continuous-control-policies-by","repo_url":"https://github.com/d3sm0/svg","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"learning-continuous-control-policies-by","repo_url":"https://github.com/opendilab/DI-engine","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"continuous-control","task_name":"Continuous Control"},{"task_slug":"continuous-control","task_name":"continuous-control"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1510.09142","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}