{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:MP6AYQ3BW63DABYVP6F2MEFUNR","short_pith_number":"pith:MP6AYQ3B","schema_version":"1.0","canonical_sha256":"63fc0c4361b7b63007157f8ba610b46c42944e3cb5c6ef63e15f748fd5253f3e","source":{"kind":"arxiv","id":"2405.12493","version":2},"attestation_state":"computed","paper":{"title":"Visualizing, Rethinking, and Mining the Loss Landscape of Deep Neural Networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"De-Chuan Zhan, Lan Li, Xin-Chun Li, Yichu Xu","submitted_at":"2024-05-21T04:30:09Z","abstract_excerpt":"The loss landscape of deep neural networks (DNNs) is commonly considered complex and wildly fluctuated. However, an interesting observation is that the loss surfaces plotted along Gaussian noise directions are almost v-basin ones with the perturbed model lying on the basin. This motivates us to rethink whether the 1D or 2D subspace could cover more complex local geometry structures, and how to mine the corresponding perturbation directions. This paper systematically and gradually categorizes the 1D curves from simple to complex, including v-basin, v-side, w-basin, w-peak, and vvv-basin curves."},"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":"2405.12493","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-05-21T04:30:09Z","cross_cats_sorted":[],"title_canon_sha256":"0449dd51997a4d116a50c70e9fb3938d1a31aae80cbac85b389f88ab28beba6e","abstract_canon_sha256":"1e895b3c3914a615ce62c4328784f366a2612726cfc56a380590f51895782574"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:16:49.187560Z","signature_b64":"8yqrLVd+/6f2oKLjFwXvHGF0agq7RBYV7T6Bu6z893FV4q/E4GBbobMMcntX9mUWQGEHgzGeHM3nWVTouXpJBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"63fc0c4361b7b63007157f8ba610b46c42944e3cb5c6ef63e15f748fd5253f3e","last_reissued_at":"2026-07-05T11:16:49.187066Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:16:49.187066Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Visualizing, Rethinking, and Mining the Loss Landscape of Deep Neural Networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"De-Chuan Zhan, Lan Li, Xin-Chun Li, Yichu Xu","submitted_at":"2024-05-21T04:30:09Z","abstract_excerpt":"The loss landscape of deep neural networks (DNNs) is commonly considered complex and wildly fluctuated. However, an interesting observation is that the loss surfaces plotted along Gaussian noise directions are almost v-basin ones with the perturbed model lying on the basin. This motivates us to rethink whether the 1D or 2D subspace could cover more complex local geometry structures, and how to mine the corresponding perturbation directions. This paper systematically and gradually categorizes the 1D curves from simple to complex, including v-basin, v-side, w-basin, w-peak, and vvv-basin curves."},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.12493","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/2405.12493/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":"2405.12493","created_at":"2026-07-05T11:16:49.187124+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.12493v2","created_at":"2026-07-05T11:16:49.187124+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.12493","created_at":"2026-07-05T11:16:49.187124+00:00"},{"alias_kind":"pith_short_12","alias_value":"MP6AYQ3BW63D","created_at":"2026-07-05T11:16:49.187124+00:00"},{"alias_kind":"pith_short_16","alias_value":"MP6AYQ3BW63DABYV","created_at":"2026-07-05T11:16:49.187124+00:00"},{"alias_kind":"pith_short_8","alias_value":"MP6AYQ3B","created_at":"2026-07-05T11:16:49.187124+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2411.16914","citing_title":"Curvature in the Looking-Glass: Optimal Methods to Exploit Curvature of Expectation in the Loss Landscape","ref_index":22,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MP6AYQ3BW63DABYVP6F2MEFUNR","json":"https://pith.science/pith/MP6AYQ3BW63DABYVP6F2MEFUNR.json","graph_json":"https://pith.science/api/pith-number/MP6AYQ3BW63DABYVP6F2MEFUNR/graph.json","events_json":"https://pith.science/api/pith-number/MP6AYQ3BW63DABYVP6F2MEFUNR/events.json","paper":"https://pith.science/paper/MP6AYQ3B"},"agent_actions":{"view_html":"https://pith.science/pith/MP6AYQ3BW63DABYVP6F2MEFUNR","download_json":"https://pith.science/pith/MP6AYQ3BW63DABYVP6F2MEFUNR.json","view_paper":"https://pith.science/paper/MP6AYQ3B","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.12493&json=true","fetch_graph":"https://pith.science/api/pith-number/MP6AYQ3BW63DABYVP6F2MEFUNR/graph.json","fetch_events":"https://pith.science/api/pith-number/MP6AYQ3BW63DABYVP6F2MEFUNR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MP6AYQ3BW63DABYVP6F2MEFUNR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MP6AYQ3BW63DABYVP6F2MEFUNR/action/storage_attestation","attest_author":"https://pith.science/pith/MP6AYQ3BW63DABYVP6F2MEFUNR/action/author_attestation","sign_citation":"https://pith.science/pith/MP6AYQ3BW63DABYVP6F2MEFUNR/action/citation_signature","submit_replication":"https://pith.science/pith/MP6AYQ3BW63DABYVP6F2MEFUNR/action/replication_record"}},"created_at":"2026-07-05T11:16:49.187124+00:00","updated_at":"2026-07-05T11:16:49.187124+00:00"}