{"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/exploratory-gradient-boosting-for","title":"Exploratory Gradient Boosting for Reinforcement Learning in Complex Domains","arxiv_id":"1603.04119","date":"2016-03-14","proceeding":null,"authors":["David Abel","Alekh Agarwal","Fernando Diaz","Akshay Krishnamurthy","Robert E. Schapire"],"abstract":"High-dimensional observations and complex real-world dynamics present major\nchallenges in reinforcement learning for both function approximation and\nexploration. We address both of these challenges with two complementary\ntechniques: First, we develop a gradient-boosting style, non-parametric\nfunction approximator for learning on $Q$-function residuals. And second, we\npropose an exploration strategy inspired by the principles of state abstraction\nand information acquisition under uncertainty. We demonstrate the empirical\neffectiveness of these techniques, first, as a preliminary check, on two\nstandard tasks (Blackjack and $n$-Chain), and then on two much larger and more\nrealistic tasks with high-dimensional observation spaces. Specifically, we\nintroduce two benchmarks built within the game Minecraft where the observations\nare pixel arrays of the agent's visual field. A combination of our two\nalgorithmic techniques performs competitively on the standard\nreinforcement-learning tasks while consistently and substantially outperforming\nbaselines on the two tasks with high-dimensional observation spaces. The new\nfunction approximator, exploration strategy, and evaluation benchmarks are each\nof independent interest in the pursuit of reinforcement-learning methods that\nscale to real-world domains.","url_abs":"http://arxiv.org/abs/1603.04119v1","url_pdf":"http://arxiv.org/pdf/1603.04119v1.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":"exploratory-gradient-boosting-for","repo_url":"https://github.com/wattlebirdaz/geql","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"minecraft","task_name":"Minecraft"},{"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":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1603.04119","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}