{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:YC4GXNLZQYSWY3RLD4NP3OHOW6","short_pith_number":"pith:YC4GXNLZ","schema_version":"1.0","canonical_sha256":"c0b86bb57986256c6e2b1f1afdb8eeb7b1550347b3c3b850de9ed01e7853f7a0","source":{"kind":"arxiv","id":"2302.03390","version":5},"attestation_state":"computed","paper":{"title":"Learning Discretized Neural Networks under Ricci Flow","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.IT","cs.NE","math.IT"],"primary_cat":"cs.LG","authors_text":"Guang Dai, Hanwen Chen, Ivor W. Tsang, Jun Chen, Mengmeng Wang, Yong Liu","submitted_at":"2023-02-07T10:51:53Z","abstract_excerpt":"In this paper, we study Discretized Neural Networks (DNNs) composed of low-precision weights and activations, which suffer from either infinite or zero gradients due to the non-differentiable discrete function during training. Most training-based DNNs in such scenarios employ the standard Straight-Through Estimator (STE) to approximate the gradient w.r.t. discrete values. However, the use of STE introduces the problem of gradient mismatch, arising from perturbations in the approximated gradient. To address this problem, this paper reveals that this mismatch can be interpreted as a metric pertu"},"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":"2302.03390","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-02-07T10:51:53Z","cross_cats_sorted":["cs.IT","cs.NE","math.IT"],"title_canon_sha256":"1633b57280467521b526098583c7c500efa02c1625d0231402b9458f3e775924","abstract_canon_sha256":"0767aae11c8219043678467f791fd0ed89f783a37b60b0a21ed41b013bced3ac"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:51:27.584582Z","signature_b64":"rMSVPFamv+XcRtqYqsNAHU4/kiLd+K7Ih9mkFGq1+IxbxC/FpyTL9hxJErb8d289YXqbIJZFChA+FJZoKQRuBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c0b86bb57986256c6e2b1f1afdb8eeb7b1550347b3c3b850de9ed01e7853f7a0","last_reissued_at":"2026-07-05T09:51:27.584155Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:51:27.584155Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning Discretized Neural Networks under Ricci Flow","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.IT","cs.NE","math.IT"],"primary_cat":"cs.LG","authors_text":"Guang Dai, Hanwen Chen, Ivor W. Tsang, Jun Chen, Mengmeng Wang, Yong Liu","submitted_at":"2023-02-07T10:51:53Z","abstract_excerpt":"In this paper, we study Discretized Neural Networks (DNNs) composed of low-precision weights and activations, which suffer from either infinite or zero gradients due to the non-differentiable discrete function during training. Most training-based DNNs in such scenarios employ the standard Straight-Through Estimator (STE) to approximate the gradient w.r.t. discrete values. However, the use of STE introduces the problem of gradient mismatch, arising from perturbations in the approximated gradient. To address this problem, this paper reveals that this mismatch can be interpreted as a metric pertu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2302.03390","kind":"arxiv","version":5},"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/2302.03390/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":"2302.03390","created_at":"2026-07-05T09:51:27.584216+00:00"},{"alias_kind":"arxiv_version","alias_value":"2302.03390v5","created_at":"2026-07-05T09:51:27.584216+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2302.03390","created_at":"2026-07-05T09:51:27.584216+00:00"},{"alias_kind":"pith_short_12","alias_value":"YC4GXNLZQYSW","created_at":"2026-07-05T09:51:27.584216+00:00"},{"alias_kind":"pith_short_16","alias_value":"YC4GXNLZQYSWY3RL","created_at":"2026-07-05T09:51:27.584216+00:00"},{"alias_kind":"pith_short_8","alias_value":"YC4GXNLZ","created_at":"2026-07-05T09:51:27.584216+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2501.03471","citing_title":"Hyperbolic Binary Neural Network","ref_index":13,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YC4GXNLZQYSWY3RLD4NP3OHOW6","json":"https://pith.science/pith/YC4GXNLZQYSWY3RLD4NP3OHOW6.json","graph_json":"https://pith.science/api/pith-number/YC4GXNLZQYSWY3RLD4NP3OHOW6/graph.json","events_json":"https://pith.science/api/pith-number/YC4GXNLZQYSWY3RLD4NP3OHOW6/events.json","paper":"https://pith.science/paper/YC4GXNLZ"},"agent_actions":{"view_html":"https://pith.science/pith/YC4GXNLZQYSWY3RLD4NP3OHOW6","download_json":"https://pith.science/pith/YC4GXNLZQYSWY3RLD4NP3OHOW6.json","view_paper":"https://pith.science/paper/YC4GXNLZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2302.03390&json=true","fetch_graph":"https://pith.science/api/pith-number/YC4GXNLZQYSWY3RLD4NP3OHOW6/graph.json","fetch_events":"https://pith.science/api/pith-number/YC4GXNLZQYSWY3RLD4NP3OHOW6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YC4GXNLZQYSWY3RLD4NP3OHOW6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YC4GXNLZQYSWY3RLD4NP3OHOW6/action/storage_attestation","attest_author":"https://pith.science/pith/YC4GXNLZQYSWY3RLD4NP3OHOW6/action/author_attestation","sign_citation":"https://pith.science/pith/YC4GXNLZQYSWY3RLD4NP3OHOW6/action/citation_signature","submit_replication":"https://pith.science/pith/YC4GXNLZQYSWY3RLD4NP3OHOW6/action/replication_record"}},"created_at":"2026-07-05T09:51:27.584216+00:00","updated_at":"2026-07-05T09:51:27.584216+00:00"}