{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:IK5426IJO2VHV37NH3OXQKLIXY","short_pith_number":"pith:IK5426IJ","schema_version":"1.0","canonical_sha256":"42bbcd790976aa7aefed3edd782968be3739bbd59f00832b55392bc4591409c7","source":{"kind":"arxiv","id":"2403.18494","version":1},"attestation_state":"computed","paper":{"title":"Learning in PINNs: Phase transition, total diffusion, and generalization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"George Em Karniadakis, Juan Diego Toscano, Nikolaos Stergiopulos, Sokratis J. Anagnostopoulos","submitted_at":"2024-03-27T12:10:30Z","abstract_excerpt":"We investigate the learning dynamics of fully-connected neural networks through the lens of gradient signal-to-noise ratio (SNR), examining the behavior of first-order optimizers like Adam in non-convex objectives. By interpreting the drift/diffusion phases in the information bottleneck theory, focusing on gradient homogeneity, we identify a third phase termed ``total diffusion\", characterized by equilibrium in the learning rates and homogeneous gradients. This phase is marked by an abrupt SNR increase, uniform residuals across the sample space and the most rapid training convergence. We propo"},"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":"2403.18494","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-03-27T12:10:30Z","cross_cats_sorted":[],"title_canon_sha256":"999484cbf9cf32f4e0fd62f678642862e9dcf3b9d876287044dfe2c3ba0ef8a2","abstract_canon_sha256":"defc93f9b99f15fe23987f0458b1a5c7460ba4c4b500e0bc28350cd6a1d33477"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:01:23.557239Z","signature_b64":"R/nFF25cCHQOr0UhaEsP5gbPPI/RG9FYj5JU3izwqj/v9AAYPmHMKxbXAk5gaI2EXZ1LvB+7s16Xdqltq87iBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"42bbcd790976aa7aefed3edd782968be3739bbd59f00832b55392bc4591409c7","last_reissued_at":"2026-07-05T08:01:23.556766Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:01:23.556766Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning in PINNs: Phase transition, total diffusion, and generalization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"George Em Karniadakis, Juan Diego Toscano, Nikolaos Stergiopulos, Sokratis J. Anagnostopoulos","submitted_at":"2024-03-27T12:10:30Z","abstract_excerpt":"We investigate the learning dynamics of fully-connected neural networks through the lens of gradient signal-to-noise ratio (SNR), examining the behavior of first-order optimizers like Adam in non-convex objectives. By interpreting the drift/diffusion phases in the information bottleneck theory, focusing on gradient homogeneity, we identify a third phase termed ``total diffusion\", characterized by equilibrium in the learning rates and homogeneous gradients. This phase is marked by an abrupt SNR increase, uniform residuals across the sample space and the most rapid training convergence. We propo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.18494","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/2403.18494/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":"2403.18494","created_at":"2026-07-05T08:01:23.556829+00:00"},{"alias_kind":"arxiv_version","alias_value":"2403.18494v1","created_at":"2026-07-05T08:01:23.556829+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.18494","created_at":"2026-07-05T08:01:23.556829+00:00"},{"alias_kind":"pith_short_12","alias_value":"IK5426IJO2VH","created_at":"2026-07-05T08:01:23.556829+00:00"},{"alias_kind":"pith_short_16","alias_value":"IK5426IJO2VHV37N","created_at":"2026-07-05T08:01:23.556829+00:00"},{"alias_kind":"pith_short_8","alias_value":"IK5426IJ","created_at":"2026-07-05T08:01:23.556829+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.07020","citing_title":"Learning Turbulence Closures with Physics-Informed Neural Networks for the Rayleigh-Taylor Transition to Turbulence","ref_index":74,"is_internal_anchor":true},{"citing_arxiv_id":"2604.07075","citing_title":"Estimating bottom topography in shallow water flows","ref_index":1,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IK5426IJO2VHV37NH3OXQKLIXY","json":"https://pith.science/pith/IK5426IJO2VHV37NH3OXQKLIXY.json","graph_json":"https://pith.science/api/pith-number/IK5426IJO2VHV37NH3OXQKLIXY/graph.json","events_json":"https://pith.science/api/pith-number/IK5426IJO2VHV37NH3OXQKLIXY/events.json","paper":"https://pith.science/paper/IK5426IJ"},"agent_actions":{"view_html":"https://pith.science/pith/IK5426IJO2VHV37NH3OXQKLIXY","download_json":"https://pith.science/pith/IK5426IJO2VHV37NH3OXQKLIXY.json","view_paper":"https://pith.science/paper/IK5426IJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2403.18494&json=true","fetch_graph":"https://pith.science/api/pith-number/IK5426IJO2VHV37NH3OXQKLIXY/graph.json","fetch_events":"https://pith.science/api/pith-number/IK5426IJO2VHV37NH3OXQKLIXY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IK5426IJO2VHV37NH3OXQKLIXY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IK5426IJO2VHV37NH3OXQKLIXY/action/storage_attestation","attest_author":"https://pith.science/pith/IK5426IJO2VHV37NH3OXQKLIXY/action/author_attestation","sign_citation":"https://pith.science/pith/IK5426IJO2VHV37NH3OXQKLIXY/action/citation_signature","submit_replication":"https://pith.science/pith/IK5426IJO2VHV37NH3OXQKLIXY/action/replication_record"}},"created_at":"2026-07-05T08:01:23.556829+00:00","updated_at":"2026-07-05T08:01:23.556829+00:00"}