{"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/uncertainty-based-offline-reinforcement","title":"Uncertainty-Based Offline Reinforcement Learning with Diversified Q-Ensemble","arxiv_id":"2110.01548","date":"2021-10-04","proceeding":"NeurIPS 2021 12","authors":["Gaon An","Seungyong Moon","Jang-Hyun Kim","Hyun Oh Song"],"abstract":"Offline reinforcement learning (offline RL), which aims to find an optimal policy from a previously collected static dataset, bears algorithmic difficulties due to function approximation errors from out-of-distribution (OOD) data points. To this end, offline RL algorithms adopt either a constraint or a penalty term that explicitly guides the policy to stay close to the given dataset. However, prior methods typically require accurate estimation of the behavior policy or sampling from OOD data points, which themselves can be a non-trivial problem. Moreover, these methods under-utilize the generalization ability of deep neural networks and often fall into suboptimal solutions too close to the given dataset. In this work, we propose an uncertainty-based offline RL method that takes into account the confidence of the Q-value prediction and does not require any estimation or sampling of the data distribution. We show that the clipped Q-learning, a technique widely used in online RL, can be leveraged to successfully penalize OOD data points with high prediction uncertainties. Surprisingly, we find that it is possible to substantially outperform existing offline RL methods on various tasks by simply increasing the number of Q-networks along with the clipped Q-learning. Based on this observation, we propose an ensemble-diversified actor-critic algorithm that reduces the number of required ensemble networks down to a tenth compared to the naive ensemble while achieving state-of-the-art performance on most of the D4RL benchmarks considered.","url_abs":"https://arxiv.org/abs/2110.01548v2","url_pdf":"https://arxiv.org/pdf/2110.01548v2.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":"uncertainty-based-offline-reinforcement","repo_url":"https://github.com/corl-team/CORL","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"jax","reach":null},{"paper_slug":"uncertainty-based-offline-reinforcement","repo_url":"https://github.com/howuhh/sac-n-jax","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"jax","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"uncertainty-based-offline-reinforcement","repo_url":"https://github.com/snu-mllab/edac","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"uncertainty-based-offline-reinforcement","repo_url":"https://github.com/yihaosun1124/OfflineRL-Kit","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"uncertainty-based-offline-reinforcement","repo_url":"https://github.com/zzmtsvv/rl_task","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"adroid-door-cloned","task_name":"Adroid door-cloned"},{"task_slug":"adroid-door-human","task_name":"Adroid door-human"},{"task_slug":"adroid-hammer-cloned","task_name":"Adroid hammer-cloned"},{"task_slug":"adroid-hammer-human","task_name":"Adroid hammer-human"},{"task_slug":"adroid-pen-cloned","task_name":"Adroid pen-cloned"},{"task_slug":"adroid-pen-human","task_name":"Adroid pen-human"},{"task_slug":"adroid-relocate-cloned","task_name":"Adroid relocate-cloned"},{"task_slug":"adroid-relocate-human","task_name":"Adroid relocate-human"},{"task_slug":"d4rl","task_name":"D4RL"},{"task_slug":"gym-halfcheetah-expert","task_name":"Gym halfcheetah-expert"},{"task_slug":"gym-halfcheetah-full-replay","task_name":"Gym halfcheetah-full-replay"},{"task_slug":"gym-halfcheetah-medium","task_name":"Gym halfcheetah-medium"},{"task_slug":"gym-halfcheetah-medium-expert","task_name":"Gym halfcheetah-medium-expert"},{"task_slug":"gym-halfcheetah-medium-replay","task_name":"Gym halfcheetah-medium-replay"},{"task_slug":"gym-halfcheetah-random","task_name":"Gym halfcheetah-random"},{"task_slug":"offline-rl","task_name":"Offline RL"},{"task_slug":"q-learning","task_name":"Q-Learning"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"},{"task_slug":"value-prediction","task_name":"Value prediction"},{"task_slug":"reinforcement-learning-2","task_name":"reinforcement-learning"}],"methods":[{"method_slug":"q-learning","method_name":"Q-Learning"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2110.01548","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2110.01548"}},"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/yihaosun1124/OfflineRL-Kit","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/corl-team/CORL","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/zzmtsvv/rl_task","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/snu-mllab/edac","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/howuhh/sac-n-jax","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"ran":12,"ran_draft_wrong":1,"unverified":8},"by_repo_kind":{"listed":{"samples":21,"ran":13,"repositories":5}},"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":6,"samples":[{"code_sha256_prefix":"7f11d5ce17556f04","entry":"Actor","repo":"corl-team/CORL","repo_kind":"listed","path":"algorithms/offline/edac.py","file_url":"https://github.com/corl-team/CORL/blob/HEAD/algorithms/offline/edac.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"deterministic","behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"7f11d5ce17556f04"}},{"code_sha256_prefix":"241911ed91897454","entry":"DeterministicActor","repo":"zzmtsvv/rl_task","repo_kind":"listed","path":"doge/doge.py","file_url":"https://github.com/zzmtsvv/rl_task/blob/HEAD/doge/doge.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"241911ed91897454"}},{"code_sha256_prefix":"ab927c7eea834816","entry":"Distance","repo":"zzmtsvv/rl_task","repo_kind":"listed","path":"doge/doge.py","file_url":"https://github.com/zzmtsvv/rl_task/blob/HEAD/doge/doge.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"ab927c7eea834816"}},{"code_sha256_prefix":"f5dc603591d69cbd","entry":"EDACPolicy","repo":"yihaosun1124/OfflineRL-Kit","repo_kind":"listed","path":"offlinerlkit/policy/model_free/edac.py","file_url":"https://github.com/yihaosun1124/OfflineRL-Kit/blob/HEAD/offlinerlkit/policy/model_free/edac.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"f5dc603591d69cbd"}},{"code_sha256_prefix":"b4a9de39db5c3e24","entry":"EnsembleCritic","repo":"howuhh/sac-n-jax","repo_kind":"listed","path":"sac_n_jax_flax.py","file_url":"https://github.com/howuhh/sac-n-jax/blob/HEAD/sac_n_jax_flax.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"b4a9de39db5c3e24"}},{"code_sha256_prefix":"31807722c6d0ebca","entry":"EnsembledCritic","repo":"zzmtsvv/rl_task","repo_kind":"listed","path":"doge/doge.py","file_url":"https://github.com/zzmtsvv/rl_task/blob/HEAD/doge/doge.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"31807722c6d0ebca"}},{"code_sha256_prefix":"5e8059b5878ab4e3","entry":"EnsembledLinear","repo":"zzmtsvv/rl_task","repo_kind":"listed","path":"doge/doge.py","file_url":"https://github.com/zzmtsvv/rl_task/blob/HEAD/doge/doge.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"5e8059b5878ab4e3"}},{"code_sha256_prefix":"97f6ff89661513fb","entry":"ParallelizedEnsemble","repo":"snu-mllab/edac","repo_kind":"listed","path":"lifelong_rl/models/networks.py","file_url":"https://github.com/snu-mllab/edac/blob/HEAD/lifelong_rl/models/networks.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"97f6ff89661513fb"}},{"code_sha256_prefix":"765f2c4c8cb57eab","entry":"ParallelizedLayer","repo":"snu-mllab/edac","repo_kind":"listed","path":"lifelong_rl/models/networks.py","file_url":"https://github.com/snu-mllab/edac/blob/HEAD/lifelong_rl/models/networks.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"765f2c4c8cb57eab"}},{"code_sha256_prefix":"832ae1f434765b20","entry":"VectorizedCritic","repo":"corl-team/CORL","repo_kind":"listed","path":"algorithms/offline/edac.py","file_url":"https://github.com/corl-team/CORL/blob/HEAD/algorithms/offline/edac.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"832ae1f434765b20"}},{"code_sha256_prefix":"0f7fd096ae2490f9","entry":"VectorizedLinear","repo":"corl-team/CORL","repo_kind":"listed","path":"algorithms/offline/edac.py","file_url":"https://github.com/corl-team/CORL/blob/HEAD/algorithms/offline/edac.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"0f7fd096ae2490f9"}},{"code_sha256_prefix":"527759337db08777","entry":"doge_config","repo":"zzmtsvv/rl_task","repo_kind":"listed","path":"doge/doge.py","file_url":"https://github.com/zzmtsvv/rl_task/blob/HEAD/doge/doge.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"527759337db08777"}},{"code_sha256_prefix":"b5491b8fa060cb74","entry":"uniform_init","repo":"howuhh/sac-n-jax","repo_kind":"listed","path":"sac_n_jax_flax.py","file_url":"https://github.com/howuhh/sac-n-jax/blob/HEAD/sac_n_jax_flax.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":"b5491b8fa060cb74"}},{"code_sha256_prefix":"5794e64ed17e773b","entry":"Critic","repo":"howuhh/sac-n-jax","repo_kind":"listed","path":"sac_n_jax_flax.py","file_url":"https://github.com/howuhh/sac-n-jax/blob/HEAD/sac_n_jax_flax.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"5794e64ed17e773b"}},{"code_sha256_prefix":"deb27526958aa790","entry":"DOGE","repo":"zzmtsvv/rl_task","repo_kind":"listed","path":"doge/doge.py","file_url":"https://github.com/zzmtsvv/rl_task/blob/HEAD/doge/doge.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"deb27526958aa790"}},{"code_sha256_prefix":"0f2060e11b06334a","entry":"EDAC","repo":"corl-team/CORL","repo_kind":"listed","path":"algorithms/offline/edac.py","file_url":"https://github.com/corl-team/CORL/blob/HEAD/algorithms/offline/edac.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"0f2060e11b06334a"}},{"code_sha256_prefix":"1d1f97e963885eb0","entry":"from_numpy","repo":"snu-mllab/edac","repo_kind":"listed","path":"lifelong_rl/models/networks.py","file_url":"https://github.com/snu-mllab/edac/blob/HEAD/lifelong_rl/models/networks.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"1d1f97e963885eb0"}},{"code_sha256_prefix":"78a13fda3d5531a2","entry":"ones","repo":"snu-mllab/edac","repo_kind":"listed","path":"lifelong_rl/models/networks.py","file_url":"https://github.com/snu-mllab/edac/blob/HEAD/lifelong_rl/models/networks.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"78a13fda3d5531a2"}},{"code_sha256_prefix":"826e9009d90a358b","entry":"randn","repo":"snu-mllab/edac","repo_kind":"listed","path":"lifelong_rl/models/networks.py","file_url":"https://github.com/snu-mllab/edac/blob/HEAD/lifelong_rl/models/networks.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"826e9009d90a358b"}},{"code_sha256_prefix":"12fd41e242a13169","entry":"soft_update","repo":"corl-team/CORL","repo_kind":"listed","path":"algorithms/offline/edac.py","file_url":"https://github.com/corl-team/CORL/blob/HEAD/algorithms/offline/edac.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"12fd41e242a13169"}},{"code_sha256_prefix":"662ecb7bbf9644d8","entry":"zeros","repo":"snu-mllab/edac","repo_kind":"listed","path":"lifelong_rl/models/networks.py","file_url":"https://github.com/snu-mllab/edac/blob/HEAD/lifelong_rl/models/networks.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"662ecb7bbf9644d8"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}