{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:J3I6O3QA2I25VXNH4HPCULYHEI","short_pith_number":"pith:J3I6O3QA","schema_version":"1.0","canonical_sha256":"4ed1e76e00d235dadda7e1de2a2f07222bd7cfbfa8e72eceb0e525782ad67e13","source":{"kind":"arxiv","id":"2607.26059","version":1},"attestation_state":"computed","paper":{"title":"Emergent Sparsity in Frozen Random CNN Feature Extractors for Deep Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.NE"],"primary_cat":"cs.LG","authors_text":"Scott M. Norton","submitted_at":"2026-05-15T00:01:05Z","abstract_excerpt":"We report a striking phenomenon: deep reinforcement learning agents trained with frozen, randomly initialized CNN feature extractors spontaneously develop extremely sparse fully-connected representations, without any sparsity-inducing objective. In the first fully-connected layer (FC1, $3{,}136 \\to 64$), agents compress task-relevant information through as few as 1-3 neurons out of 64 for deterministic Pong (5-11 for stochastic Pong), while trainable CNNs activate 55-64 neurons under matched conditions. We establish four principal findings. First, FC1 sparsity scales with task complexity: 1-11"},"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":"2607.26059","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-15T00:01:05Z","cross_cats_sorted":["cs.AI","cs.NE"],"title_canon_sha256":"5508169159b3d2ecccc281053b997cc05c5014a97a9b84b5f5bce6d889440cdf","abstract_canon_sha256":"fc8fba05a6196fb64f832fc035955e67a5a36f208bd9e266ff35ca7faa0e4df0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4ed1e76e00d235dadda7e1de2a2f07222bd7cfbfa8e72eceb0e525782ad67e13","last_reissued_at":"2026-07-30T00:08:16.321452Z","signature_status":"unsigned_v0","first_computed_at":"2026-07-30T00:08:16.321452Z"},"graph_snapshot":{"paper":{"title":"Emergent Sparsity in Frozen Random CNN Feature Extractors for Deep Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.NE"],"primary_cat":"cs.LG","authors_text":"Scott M. Norton","submitted_at":"2026-05-15T00:01:05Z","abstract_excerpt":"We report a striking phenomenon: deep reinforcement learning agents trained with frozen, randomly initialized CNN feature extractors spontaneously develop extremely sparse fully-connected representations, without any sparsity-inducing objective. In the first fully-connected layer (FC1, $3{,}136 \\to 64$), agents compress task-relevant information through as few as 1-3 neurons out of 64 for deterministic Pong (5-11 for stochastic Pong), while trainable CNNs activate 55-64 neurons under matched conditions. We establish four principal findings. First, FC1 sparsity scales with task complexity: 1-11"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.26059","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":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2607.26059/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":"2607.26059","created_at":"2026-07-30T00:08:16.325019+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.26059v1","created_at":"2026-07-30T00:08:16.325019+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.26059","created_at":"2026-07-30T00:08:16.325019+00:00"},{"alias_kind":"pith_short_12","alias_value":"J3I6O3QA2I25","created_at":"2026-07-30T00:08:16.325019+00:00"},{"alias_kind":"pith_short_16","alias_value":"J3I6O3QA2I25VXNH","created_at":"2026-07-30T00:08:16.325019+00:00"},{"alias_kind":"pith_short_8","alias_value":"J3I6O3QA","created_at":"2026-07-30T00:08:16.325019+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/J3I6O3QA2I25VXNH4HPCULYHEI","json":"https://pith.science/pith/J3I6O3QA2I25VXNH4HPCULYHEI.json","graph_json":"https://pith.science/api/pith-number/J3I6O3QA2I25VXNH4HPCULYHEI/graph.json","events_json":"https://pith.science/api/pith-number/J3I6O3QA2I25VXNH4HPCULYHEI/events.json","paper":"https://pith.science/paper/J3I6O3QA"},"agent_actions":{"view_html":"https://pith.science/pith/J3I6O3QA2I25VXNH4HPCULYHEI","download_json":"https://pith.science/pith/J3I6O3QA2I25VXNH4HPCULYHEI.json","view_paper":"https://pith.science/paper/J3I6O3QA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.26059&json=true","fetch_graph":"https://pith.science/api/pith-number/J3I6O3QA2I25VXNH4HPCULYHEI/graph.json","fetch_events":"https://pith.science/api/pith-number/J3I6O3QA2I25VXNH4HPCULYHEI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/J3I6O3QA2I25VXNH4HPCULYHEI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/J3I6O3QA2I25VXNH4HPCULYHEI/action/storage_attestation","attest_author":"https://pith.science/pith/J3I6O3QA2I25VXNH4HPCULYHEI/action/author_attestation","sign_citation":"https://pith.science/pith/J3I6O3QA2I25VXNH4HPCULYHEI/action/citation_signature","submit_replication":"https://pith.science/pith/J3I6O3QA2I25VXNH4HPCULYHEI/action/replication_record"}},"created_at":"2026-07-30T00:08:16.325019+00:00","updated_at":"2026-07-30T00:08:16.325019+00:00"}