{"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/illuminating-generalization-in-deep","title":"Illuminating Generalization in Deep Reinforcement Learning through Procedural Level Generation","arxiv_id":"1806.10729","date":"2018-06-28","proceeding":null,"authors":["Niels Justesen","Ruben Rodriguez Torrado","Philip Bontrager","Ahmed Khalifa","Julian Togelius","Sebastian Risi"],"abstract":"Deep reinforcement learning (RL) has shown impressive results in a variety of\ndomains, learning directly from high-dimensional sensory streams. However, when\nneural networks are trained in a fixed environment, such as a single level in a\nvideo game, they will usually overfit and fail to generalize to new levels.\nWhen RL models overfit, even slight modifications to the environment can result\nin poor agent performance. This paper explores how procedurally generated\nlevels during training can increase generality. We show that for some games\nprocedural level generation enables generalization to new levels within the\nsame distribution. Additionally, it is possible to achieve better performance\nwith less data by manipulating the difficulty of the levels in response to the\nperformance of the agent. The generality of the learned behaviors is also\nevaluated on a set of human-designed levels. The results suggest that the\nability to generalize to human-designed levels highly depends on the design of\nthe level generators. We apply dimensionality reduction and clustering\ntechniques to visualize the generators' distributions of levels and analyze to\nwhat degree they can produce levels similar to those designed by a human.","url_abs":"http://arxiv.org/abs/1806.10729v5","url_pdf":"http://arxiv.org/pdf/1806.10729v5.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":"illuminating-generalization-in-deep","repo_url":"https://github.com/njustesen/a2c_gvgai","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok","spdx":"GPL-3.0"}}],"tasks":[{"task_slug":"clustering","task_name":"Clustering"},{"task_slug":"deep-reinforcement-learning","task_name":"Deep Reinforcement Learning"},{"task_slug":"dimensionality-reduction","task_name":"Dimensionality Reduction"},{"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=1806.10729","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}