{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2020:KRHTZLJYFHIQDV7BUVRV3GQTLH","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"80980e9f40f6cdb655d5b5a159a81ebe72c29dab22601bbec577ad63e1389f16","cross_cats_sorted":["cond-mat.dis-nn","cond-mat.stat-mech","q-bio.NC","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-12-08T20:33:30Z","title_canon_sha256":"c2a724abd22932a7999fea160605e343149e9ce9e8f9aa42b108526cd576f04e"},"schema_version":"1.0","source":{"id":"2012.04728","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2012.04728","created_at":"2026-07-05T02:26:53Z"},{"alias_kind":"arxiv_version","alias_value":"2012.04728v2","created_at":"2026-07-05T02:26:53Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2012.04728","created_at":"2026-07-05T02:26:53Z"},{"alias_kind":"pith_short_12","alias_value":"KRHTZLJYFHIQ","created_at":"2026-07-05T02:26:53Z"},{"alias_kind":"pith_short_16","alias_value":"KRHTZLJYFHIQDV7B","created_at":"2026-07-05T02:26:53Z"},{"alias_kind":"pith_short_8","alias_value":"KRHTZLJY","created_at":"2026-07-05T02:26:53Z"}],"graph_snapshots":[{"event_id":"sha256:032d64b15f077bbbe2a1bec8cbaed21b060b65d7e5636e526211c99c02b00e2f","target":"graph","created_at":"2026-07-05T02:26:53Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2012.04728/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Understanding the dynamics of neural network parameters during training is one of the key challenges in building a theoretical foundation for deep learning. A central obstacle is that the motion of a network in high-dimensional parameter space undergoes discrete finite steps along complex stochastic gradients derived from real-world datasets. We circumvent this obstacle through a unifying theoretical framework based on intrinsic symmetries embedded in a network's architecture that are present for any dataset. We show that any such symmetry imposes stringent geometric constraints on gradients a","authors_text":"Daniel Kunin, Daniel L.K. Yamins, Hidenori Tanaka, Javier Sagastuy-Brena, Surya Ganguli","cross_cats":["cond-mat.dis-nn","cond-mat.stat-mech","q-bio.NC","stat.ML"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-12-08T20:33:30Z","title":"Neural Mechanics: Symmetry and Broken Conservation Laws in Deep Learning Dynamics"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2012.04728","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:0e36d136d530fffe305ad4ce0b87ec6b668d0995f7ab92eb19588d2582c95f63","target":"record","created_at":"2026-07-05T02:26:53Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"80980e9f40f6cdb655d5b5a159a81ebe72c29dab22601bbec577ad63e1389f16","cross_cats_sorted":["cond-mat.dis-nn","cond-mat.stat-mech","q-bio.NC","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-12-08T20:33:30Z","title_canon_sha256":"c2a724abd22932a7999fea160605e343149e9ce9e8f9aa42b108526cd576f04e"},"schema_version":"1.0","source":{"id":"2012.04728","kind":"arxiv","version":2}},"canonical_sha256":"544f3cad3829d101d7e1a5635d9a1359c154f1bbf15f6ec53782dee1ff822aeb","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"544f3cad3829d101d7e1a5635d9a1359c154f1bbf15f6ec53782dee1ff822aeb","first_computed_at":"2026-07-05T02:26:53.078549Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T02:26:53.078549Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"2bYZDcvucaqE2Dy+xb3F5yvgTRayUJI/v1puemSkq/bOwOPVN4VBa9JLrTjwmvzE7VA2JgP/4P3EwDDo/1+pDg==","signature_status":"signed_v1","signed_at":"2026-07-05T02:26:53.078968Z","signed_message":"canonical_sha256_bytes"},"source_id":"2012.04728","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:0e36d136d530fffe305ad4ce0b87ec6b668d0995f7ab92eb19588d2582c95f63","sha256:032d64b15f077bbbe2a1bec8cbaed21b060b65d7e5636e526211c99c02b00e2f"],"state_sha256":"e48232a0de007bb24da17dc9f7fd85c13ae763e9f951805c996da980e8ed93a2"}