{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2017:QJCG6RVIUNB4K4HU2E4ZAL2D4M","short_pith_number":"pith:QJCG6RVI","schema_version":"1.0","canonical_sha256":"82446f46a8a343c570f4d139902f43e31fb1da83564743f898a56b05e5103e69","source":{"kind":"arxiv","id":"1701.04143","version":1},"attestation_state":"computed","paper":{"title":"Vulnerability of Deep Reinforcement Learning to Policy Induction Attacks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Arslan Munir, Vahid Behzadan","submitted_at":"2017-01-16T02:39:01Z","abstract_excerpt":"Deep learning classifiers are known to be inherently vulnerable to manipulation by intentionally perturbed inputs, named adversarial examples. In this work, we establish that reinforcement learning techniques based on Deep Q-Networks (DQNs) are also vulnerable to adversarial input perturbations, and verify the transferability of adversarial examples across different DQN models. Furthermore, we present a novel class of attacks based on this vulnerability that enable policy manipulation and induction in the learning process of DQNs. We propose an attack mechanism that exploits the transferabilit"},"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":"1701.04143","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2017-01-16T02:39:01Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"95c6ebcafa4a9d663896400374e736f21ce5ba4b39dfb89ce87e8f31bfc8dfa2","abstract_canon_sha256":"906b8af8ae416bbf4bcd4665adcf4cdcf8fe1d322ba61eed4cd4f48369e67d10"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T00:52:47.742150Z","signature_b64":"fc3ElDaHeTF9vew8MR8TbKK63UMSrObAe6cbYBN8z8Ezv5Ky4bfshR9WSGyEnD43MQJgTO/DEEVK3rmBBTxaCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"82446f46a8a343c570f4d139902f43e31fb1da83564743f898a56b05e5103e69","last_reissued_at":"2026-05-18T00:52:47.741527Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T00:52:47.741527Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Vulnerability of Deep Reinforcement Learning to Policy Induction Attacks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Arslan Munir, Vahid Behzadan","submitted_at":"2017-01-16T02:39:01Z","abstract_excerpt":"Deep learning classifiers are known to be inherently vulnerable to manipulation by intentionally perturbed inputs, named adversarial examples. In this work, we establish that reinforcement learning techniques based on Deep Q-Networks (DQNs) are also vulnerable to adversarial input perturbations, and verify the transferability of adversarial examples across different DQN models. Furthermore, we present a novel class of attacks based on this vulnerability that enable policy manipulation and induction in the learning process of DQNs. We propose an attack mechanism that exploits the transferabilit"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1701.04143","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"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":"1701.04143","created_at":"2026-05-18T00:52:47.741627+00:00"},{"alias_kind":"arxiv_version","alias_value":"1701.04143v1","created_at":"2026-05-18T00:52:47.741627+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1701.04143","created_at":"2026-05-18T00:52:47.741627+00:00"},{"alias_kind":"pith_short_12","alias_value":"QJCG6RVIUNB4","created_at":"2026-05-18T12:31:39.905425+00:00"},{"alias_kind":"pith_short_16","alias_value":"QJCG6RVIUNB4K4HU","created_at":"2026-05-18T12:31:39.905425+00:00"},{"alias_kind":"pith_short_8","alias_value":"QJCG6RVI","created_at":"2026-05-18T12:31:39.905425+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2412.16213","citing_title":"AdvIRL: Reinforcement Learning-Based Adversarial Attacks on 3D NeRF Models","ref_index":3,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QJCG6RVIUNB4K4HU2E4ZAL2D4M","json":"https://pith.science/pith/QJCG6RVIUNB4K4HU2E4ZAL2D4M.json","graph_json":"https://pith.science/api/pith-number/QJCG6RVIUNB4K4HU2E4ZAL2D4M/graph.json","events_json":"https://pith.science/api/pith-number/QJCG6RVIUNB4K4HU2E4ZAL2D4M/events.json","paper":"https://pith.science/paper/QJCG6RVI"},"agent_actions":{"view_html":"https://pith.science/pith/QJCG6RVIUNB4K4HU2E4ZAL2D4M","download_json":"https://pith.science/pith/QJCG6RVIUNB4K4HU2E4ZAL2D4M.json","view_paper":"https://pith.science/paper/QJCG6RVI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1701.04143&json=true","fetch_graph":"https://pith.science/api/pith-number/QJCG6RVIUNB4K4HU2E4ZAL2D4M/graph.json","fetch_events":"https://pith.science/api/pith-number/QJCG6RVIUNB4K4HU2E4ZAL2D4M/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QJCG6RVIUNB4K4HU2E4ZAL2D4M/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QJCG6RVIUNB4K4HU2E4ZAL2D4M/action/storage_attestation","attest_author":"https://pith.science/pith/QJCG6RVIUNB4K4HU2E4ZAL2D4M/action/author_attestation","sign_citation":"https://pith.science/pith/QJCG6RVIUNB4K4HU2E4ZAL2D4M/action/citation_signature","submit_replication":"https://pith.science/pith/QJCG6RVIUNB4K4HU2E4ZAL2D4M/action/replication_record"}},"created_at":"2026-05-18T00:52:47.741627+00:00","updated_at":"2026-05-18T00:52:47.741627+00:00"}