{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:GNISDVH6CSTB2OOPGEHY5PGAVG","short_pith_number":"pith:GNISDVH6","schema_version":"1.0","canonical_sha256":"335121d4fe14a61d39cf310f8ebcc0a991f3a778ee441946d0c5a904a0d8b906","source":{"kind":"arxiv","id":"2006.10742","version":2},"attestation_state":"computed","paper":{"title":"Learning Invariant Representations for Reinforcement Learning without Reconstruction","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Amy Zhang, Roberto Calandra, Rowan McAllister, Sergey Levine, Yarin Gal","submitted_at":"2020-06-18T17:59:35Z","abstract_excerpt":"We study how representation learning can accelerate reinforcement learning from rich observations, such as images, without relying either on domain knowledge or pixel-reconstruction. Our goal is to learn representations that both provide for effective downstream control and invariance to task-irrelevant details. Bisimulation metrics quantify behavioral similarity between states in continuous MDPs, which we propose using to learn robust latent representations which encode only the task-relevant information from observations. Our method trains encoders such that distances in latent space equal b"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2006.10742","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-06-18T17:59:35Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"5e5ff642737d2bdf07fa74bc0977ccc9c561b3037b03e381f6dc044c89a81aa2","abstract_canon_sha256":"a5ac723821cb6a676c5c71cebe424405b557d833465d3d68e61423a2e655c118"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:29:53.895710Z","signature_b64":"P5eIwMannzmltJ4bUBuUCRNu51QIj3evzN9GaeQLpD7aYlwvoGqstjsf4pCNCj6mLczuE9c8c9CSXDDwBOC6AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"335121d4fe14a61d39cf310f8ebcc0a991f3a778ee441946d0c5a904a0d8b906","last_reissued_at":"2026-07-05T02:29:53.895227Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:29:53.895227Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning Invariant Representations for Reinforcement Learning without Reconstruction","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Amy Zhang, Roberto Calandra, Rowan McAllister, Sergey Levine, Yarin Gal","submitted_at":"2020-06-18T17:59:35Z","abstract_excerpt":"We study how representation learning can accelerate reinforcement learning from rich observations, such as images, without relying either on domain knowledge or pixel-reconstruction. Our goal is to learn representations that both provide for effective downstream control and invariance to task-irrelevant details. Bisimulation metrics quantify behavioral similarity between states in continuous MDPs, which we propose using to learn robust latent representations which encode only the task-relevant information from observations. Our method trains encoders such that distances in latent space equal b"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2006.10742","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2006.10742/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2006.10742","created_at":"2026-07-05T02:29:53.895289+00:00"},{"alias_kind":"arxiv_version","alias_value":"2006.10742v2","created_at":"2026-07-05T02:29:53.895289+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2006.10742","created_at":"2026-07-05T02:29:53.895289+00:00"},{"alias_kind":"pith_short_12","alias_value":"GNISDVH6CSTB","created_at":"2026-07-05T02:29:53.895289+00:00"},{"alias_kind":"pith_short_16","alias_value":"GNISDVH6CSTB2OOP","created_at":"2026-07-05T02:29:53.895289+00:00"},{"alias_kind":"pith_short_8","alias_value":"GNISDVH6","created_at":"2026-07-05T02:29:53.895289+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":13,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25556","citing_title":"BiPACE: Bisimulation-Guided Policy Optimization with Action Counterfactual Estimation for LLM Agents","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2606.25421","citing_title":"Beyond Next-Observation Prediction: Agent-Authored World Modeling for Sequential Decision Making","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2606.20411","citing_title":"Direct Advantage Estimation for Scalable and Sample-efficient Deep Reinforcement Learning","ref_index":82,"is_internal_anchor":false},{"citing_arxiv_id":"2606.11525","citing_title":"Learning Object Manipulation from Scratch via Contrastive Interaction","ref_index":71,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27681","citing_title":"Textual Belief States for World Models: Identifiable Representation Learning Under Strict Mediation","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00427","citing_title":"Topology-Aware State Abstraction with Tangle Cores for Markov Decision Processes","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2511.05963","citing_title":"Next-Latent Prediction Transformers Learn Compact World Models","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2411.05174","citing_title":"Bayesian Inverse Transition Learning: Learning Dynamics From Near-Optimal Trajectories","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2412.08893","citing_title":"Optimal Control with Natural Images: Efficient Reinforcement Learning using Overcomplete Sparse Codes","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2203.12601","citing_title":"R3M: A Universal Visual Representation for Robot Manipulation","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14304","citing_title":"Matrix-Space Reinforcement Learning for Reusing Local Transition Geometry","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.13105","citing_title":"What to Ignore, What to React: Visually Robust RL Fine-Tuning of VLA Models","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07278","citing_title":"Predictive but Not Plannable: RC-aux for Latent World Models","ref_index":46,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GNISDVH6CSTB2OOPGEHY5PGAVG","json":"https://pith.science/pith/GNISDVH6CSTB2OOPGEHY5PGAVG.json","graph_json":"https://pith.science/api/pith-number/GNISDVH6CSTB2OOPGEHY5PGAVG/graph.json","events_json":"https://pith.science/api/pith-number/GNISDVH6CSTB2OOPGEHY5PGAVG/events.json","paper":"https://pith.science/paper/GNISDVH6"},"agent_actions":{"view_html":"https://pith.science/pith/GNISDVH6CSTB2OOPGEHY5PGAVG","download_json":"https://pith.science/pith/GNISDVH6CSTB2OOPGEHY5PGAVG.json","view_paper":"https://pith.science/paper/GNISDVH6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2006.10742&json=true","fetch_graph":"https://pith.science/api/pith-number/GNISDVH6CSTB2OOPGEHY5PGAVG/graph.json","fetch_events":"https://pith.science/api/pith-number/GNISDVH6CSTB2OOPGEHY5PGAVG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GNISDVH6CSTB2OOPGEHY5PGAVG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GNISDVH6CSTB2OOPGEHY5PGAVG/action/storage_attestation","attest_author":"https://pith.science/pith/GNISDVH6CSTB2OOPGEHY5PGAVG/action/author_attestation","sign_citation":"https://pith.science/pith/GNISDVH6CSTB2OOPGEHY5PGAVG/action/citation_signature","submit_replication":"https://pith.science/pith/GNISDVH6CSTB2OOPGEHY5PGAVG/action/replication_record"}},"created_at":"2026-07-05T02:29:53.895289+00:00","updated_at":"2026-07-05T02:29:53.895289+00:00"}