{"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/count-based-exploration-with-neural-density","title":"Count-Based Exploration with Neural Density Models","arxiv_id":"1703.01310","date":"2017-03-03","proceeding":"ICML 2017 8","authors":["Georg Ostrovski","Marc G. Bellemare","Aaron van den Oord","Remi Munos"],"abstract":"Bellemare et al. (2016) introduced the notion of a pseudo-count, derived from\na density model, to generalize count-based exploration to non-tabular\nreinforcement learning. This pseudo-count was used to generate an exploration\nbonus for a DQN agent and combined with a mixed Monte Carlo update was\nsufficient to achieve state of the art on the Atari 2600 game Montezuma's\nRevenge. We consider two questions left open by their work: First, how\nimportant is the quality of the density model for exploration? Second, what\nrole does the Monte Carlo update play in exploration? We answer the first\nquestion by demonstrating the use of PixelCNN, an advanced neural density model\nfor images, to supply a pseudo-count. In particular, we examine the intrinsic\ndifficulties in adapting Bellemare et al.'s approach when assumptions about the\nmodel are violated. The result is a more practical and general algorithm\nrequiring no special apparatus. We combine PixelCNN pseudo-counts with\ndifferent agent architectures to dramatically improve the state of the art on\nseveral hard Atari games. One surprising finding is that the mixed Monte Carlo\nupdate is a powerful facilitator of exploration in the sparsest of settings,\nincluding Montezuma's Revenge.","url_abs":"http://arxiv.org/abs/1703.01310v2","url_pdf":"http://arxiv.org/pdf/1703.01310v2.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":"count-based-exploration-with-neural-density","repo_url":"https://github.com/nolisten/erl","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"atari-games","task_name":"Atari Games"},{"task_slug":"montezumas-revenge","task_name":"Montezuma's Revenge"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"}],"methods":[{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"dqn","method_name":"DQN"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"pixelcnn","method_name":"PixelCNN"},{"method_slug":"q-learning","method_name":"Q-Learning"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/atari-games-on-atari-2600-freeway","task":"Atari Games","dataset":"Atari 2600 Freeway","model":"DQN-CTS","rank_in_archive_order":20,"of":59,"metrics":{"Score":"33.0"},"uses_additional_data":false},{"leaderboard":"/sota/atari-games-on-atari-2600-freeway","task":"Atari Games","dataset":"Atari 2600 Freeway","model":"DQN-PixelCNN","rank_in_archive_order":29,"of":59,"metrics":{"Score":"31.7"},"uses_additional_data":false},{"leaderboard":"/sota/atari-games-on-atari-2600-gravitar","task":"Atari Games","dataset":"Atari 2600 Gravitar","model":"DQN-PixelCNN","rank_in_archive_order":31,"of":53,"metrics":{"Score":"498.3"},"uses_additional_data":false},{"leaderboard":"/sota/atari-games-on-atari-2600-gravitar","task":"Atari Games","dataset":"Atari 2600 Gravitar","model":"DQN-CTS","rank_in_archive_order":51,"of":53,"metrics":{"Score":"238.0"},"uses_additional_data":false},{"leaderboard":"/sota/atari-games-on-atari-2600-montezumas-revenge","task":"Atari Games","dataset":"Atari 2600 Montezuma's Revenge","model":"DQN-PixelCNN","rank_in_archive_order":9,"of":50,"metrics":{"Score":"3705.5"},"uses_additional_data":false},{"leaderboard":"/sota/atari-games-on-atari-2600-private-eye","task":"Atari Games","dataset":"Atari 2600 Private Eye","model":"DQN-PixelCNN","rank_in_archive_order":12,"of":52,"metrics":{"Score":"8358.7"},"uses_additional_data":false},{"leaderboard":"/sota/atari-games-on-atari-2600-private-eye","task":"Atari Games","dataset":"Atari 2600 Private Eye","model":"DQN-CTS","rank_in_archive_order":35,"of":52,"metrics":{"Score":"206.0"},"uses_additional_data":false},{"leaderboard":"/sota/atari-games-on-atari-2600-venture","task":"Atari Games","dataset":"Atari 2600 Venture","model":"DQN-PixelCNN","rank_in_archive_order":34,"of":55,"metrics":{"Score":"82.2"},"uses_additional_data":false},{"leaderboard":"/sota/atari-games-on-atari-2600-venture","task":"Atari Games","dataset":"Atari 2600 Venture","model":"DQN-CTS","rank_in_archive_order":38,"of":55,"metrics":{"Score":"48.0"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1703.01310","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1703.01310"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/nolisten/erl","reach":null}],"summary":{"ran_draft_wrong":1},"by_repo_kind":{"listed":{"samples":1,"ran":1,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"c05e0272ac8fe165","entry":"init_history","repo":"nolisten/erl","repo_kind":"listed","path":"dqn/agent.py","file_url":"https://github.com/nolisten/erl/blob/HEAD/dqn/agent.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"c05e0272ac8fe165"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}