{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:FLCV6VO6D6VLOTLXFMUGQU5ZHU","short_pith_number":"pith:FLCV6VO6","schema_version":"1.0","canonical_sha256":"2ac55f55de1faab74d772b286853b93d34b19c67bd587fe054fb91f77a9d0331","source":{"kind":"arxiv","id":"1905.02479","version":1},"attestation_state":"computed","paper":{"title":"P2SGrad: Refined Gradients for Optimizing Deep Face Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Hongsheng Li, Junjie Yan, Mengya Gao, Rui Zhao, Xiaogang Wang, Xiao Zhang, Yu Qiao","submitted_at":"2019-05-07T11:38:29Z","abstract_excerpt":"Cosine-based softmax losses significantly improve the performance of deep face recognition networks. However, these losses always include sensitive hyper-parameters which can make training process unstable, and it is very tricky to set suitable hyper parameters for a specific dataset. This paper addresses this challenge by directly designing the gradients for adaptively training deep neural networks. We first investigate and unify previous cosine softmax losses by analyzing their gradients. This unified view inspires us to propose a novel gradient called P2SGrad (Probability-to-Similarity Grad"},"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":"1905.02479","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2019-05-07T11:38:29Z","cross_cats_sorted":[],"title_canon_sha256":"a39ec0a28a8cb76d1749bcfa88f9089f8534b18704878362dd8593d5722332bb","abstract_canon_sha256":"c1f025b4f4301662f5705e3cacd4dfc2090df0aab2bd674261493f42af54c880"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:46:51.303351Z","signature_b64":"E2Q4XCfIujHSGhHYhEz8Ted8cY+dzi2BKQUQippHhqJR6FOey9xWhnW3EooQCwYm9GAHHyUOsaatC70j/fTnCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2ac55f55de1faab74d772b286853b93d34b19c67bd587fe054fb91f77a9d0331","last_reissued_at":"2026-05-17T23:46:51.302768Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:46:51.302768Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"P2SGrad: Refined Gradients for Optimizing Deep Face Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Hongsheng Li, Junjie Yan, Mengya Gao, Rui Zhao, Xiaogang Wang, Xiao Zhang, Yu Qiao","submitted_at":"2019-05-07T11:38:29Z","abstract_excerpt":"Cosine-based softmax losses significantly improve the performance of deep face recognition networks. However, these losses always include sensitive hyper-parameters which can make training process unstable, and it is very tricky to set suitable hyper parameters for a specific dataset. This paper addresses this challenge by directly designing the gradients for adaptively training deep neural networks. We first investigate and unify previous cosine softmax losses by analyzing their gradients. This unified view inspires us to propose a novel gradient called P2SGrad (Probability-to-Similarity Grad"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1905.02479","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":"1905.02479","created_at":"2026-05-17T23:46:51.302877+00:00"},{"alias_kind":"arxiv_version","alias_value":"1905.02479v1","created_at":"2026-05-17T23:46:51.302877+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1905.02479","created_at":"2026-05-17T23:46:51.302877+00:00"},{"alias_kind":"pith_short_12","alias_value":"FLCV6VO6D6VL","created_at":"2026-05-18T12:33:15.570797+00:00"},{"alias_kind":"pith_short_16","alias_value":"FLCV6VO6D6VLOTLX","created_at":"2026-05-18T12:33:15.570797+00:00"},{"alias_kind":"pith_short_8","alias_value":"FLCV6VO6","created_at":"2026-05-18T12:33:15.570797+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.07232","citing_title":"Towards multi-modal forgery representation learning for AI-generated video detection and localization","ref_index":21,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FLCV6VO6D6VLOTLXFMUGQU5ZHU","json":"https://pith.science/pith/FLCV6VO6D6VLOTLXFMUGQU5ZHU.json","graph_json":"https://pith.science/api/pith-number/FLCV6VO6D6VLOTLXFMUGQU5ZHU/graph.json","events_json":"https://pith.science/api/pith-number/FLCV6VO6D6VLOTLXFMUGQU5ZHU/events.json","paper":"https://pith.science/paper/FLCV6VO6"},"agent_actions":{"view_html":"https://pith.science/pith/FLCV6VO6D6VLOTLXFMUGQU5ZHU","download_json":"https://pith.science/pith/FLCV6VO6D6VLOTLXFMUGQU5ZHU.json","view_paper":"https://pith.science/paper/FLCV6VO6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1905.02479&json=true","fetch_graph":"https://pith.science/api/pith-number/FLCV6VO6D6VLOTLXFMUGQU5ZHU/graph.json","fetch_events":"https://pith.science/api/pith-number/FLCV6VO6D6VLOTLXFMUGQU5ZHU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FLCV6VO6D6VLOTLXFMUGQU5ZHU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FLCV6VO6D6VLOTLXFMUGQU5ZHU/action/storage_attestation","attest_author":"https://pith.science/pith/FLCV6VO6D6VLOTLXFMUGQU5ZHU/action/author_attestation","sign_citation":"https://pith.science/pith/FLCV6VO6D6VLOTLXFMUGQU5ZHU/action/citation_signature","submit_replication":"https://pith.science/pith/FLCV6VO6D6VLOTLXFMUGQU5ZHU/action/replication_record"}},"created_at":"2026-05-17T23:46:51.302877+00:00","updated_at":"2026-05-17T23:46:51.302877+00:00"}