{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:R3GDJ3QQ3C2ZQDFYOKDE5SKEUU","short_pith_number":"pith:R3GDJ3QQ","schema_version":"1.0","canonical_sha256":"8ecc34ee10d8b5980cb872864ec944a504ca70cab9d6bce352c38765ac3aa3e6","source":{"kind":"arxiv","id":"2002.06504","version":2},"attestation_state":"computed","paper":{"title":"Differentiable Top-k Operator with Optimal Transport","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Bo Dai, Hanjun Dai, Hongyuan Zha, Minshuo Chen, Tomas Pfister, Tuo Zhao, Wei Wei, Yujia Xie","submitted_at":"2020-02-16T04:57:52Z","abstract_excerpt":"The top-k operation, i.e., finding the k largest or smallest elements from a collection of scores, is an important model component, which is widely used in information retrieval, machine learning, and data mining. However, if the top-k operation is implemented in an algorithmic way, e.g., using bubble algorithm, the resulting model cannot be trained in an end-to-end way using prevalent gradient descent algorithms. This is because these implementations typically involve swapping indices, whose gradient cannot be computed. Moreover, the corresponding mapping from the input scores to the indicato"},"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":"2002.06504","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-02-16T04:57:52Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"b19ed6d0417e05f91c5ee1f95fe7ace65b7721387a7024aa5a406ead3c6bd11e","abstract_canon_sha256":"db70121f04fd1d5ea46f0cfd73f3cd37996a59d2d482eab971ce2ff172334d33"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:42:00.816084Z","signature_b64":"7K47ZxGiTZl+7rM3w63/+1tRScbDI7jBsQ2yWGc1zyyPsECASYg+pZA6t40jp2C78WlzUhTrYB7eXAxNdP5jCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8ecc34ee10d8b5980cb872864ec944a504ca70cab9d6bce352c38765ac3aa3e6","last_reissued_at":"2026-07-05T00:42:00.815451Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:42:00.815451Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Differentiable Top-k Operator with Optimal Transport","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Bo Dai, Hanjun Dai, Hongyuan Zha, Minshuo Chen, Tomas Pfister, Tuo Zhao, Wei Wei, Yujia Xie","submitted_at":"2020-02-16T04:57:52Z","abstract_excerpt":"The top-k operation, i.e., finding the k largest or smallest elements from a collection of scores, is an important model component, which is widely used in information retrieval, machine learning, and data mining. However, if the top-k operation is implemented in an algorithmic way, e.g., using bubble algorithm, the resulting model cannot be trained in an end-to-end way using prevalent gradient descent algorithms. This is because these implementations typically involve swapping indices, whose gradient cannot be computed. Moreover, the corresponding mapping from the input scores to the indicato"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2002.06504","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/2002.06504/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":"2002.06504","created_at":"2026-07-05T00:42:00.815527+00:00"},{"alias_kind":"arxiv_version","alias_value":"2002.06504v2","created_at":"2026-07-05T00:42:00.815527+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2002.06504","created_at":"2026-07-05T00:42:00.815527+00:00"},{"alias_kind":"pith_short_12","alias_value":"R3GDJ3QQ3C2Z","created_at":"2026-07-05T00:42:00.815527+00:00"},{"alias_kind":"pith_short_16","alias_value":"R3GDJ3QQ3C2ZQDFY","created_at":"2026-07-05T00:42:00.815527+00:00"},{"alias_kind":"pith_short_8","alias_value":"R3GDJ3QQ","created_at":"2026-07-05T00:42:00.815527+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2608.04057","citing_title":"LaPrune: Controllable Differentiable Sparsity at Million Scale","ref_index":16,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/R3GDJ3QQ3C2ZQDFYOKDE5SKEUU","json":"https://pith.science/pith/R3GDJ3QQ3C2ZQDFYOKDE5SKEUU.json","graph_json":"https://pith.science/api/pith-number/R3GDJ3QQ3C2ZQDFYOKDE5SKEUU/graph.json","events_json":"https://pith.science/api/pith-number/R3GDJ3QQ3C2ZQDFYOKDE5SKEUU/events.json","paper":"https://pith.science/paper/R3GDJ3QQ"},"agent_actions":{"view_html":"https://pith.science/pith/R3GDJ3QQ3C2ZQDFYOKDE5SKEUU","download_json":"https://pith.science/pith/R3GDJ3QQ3C2ZQDFYOKDE5SKEUU.json","view_paper":"https://pith.science/paper/R3GDJ3QQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2002.06504&json=true","fetch_graph":"https://pith.science/api/pith-number/R3GDJ3QQ3C2ZQDFYOKDE5SKEUU/graph.json","fetch_events":"https://pith.science/api/pith-number/R3GDJ3QQ3C2ZQDFYOKDE5SKEUU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/R3GDJ3QQ3C2ZQDFYOKDE5SKEUU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/R3GDJ3QQ3C2ZQDFYOKDE5SKEUU/action/storage_attestation","attest_author":"https://pith.science/pith/R3GDJ3QQ3C2ZQDFYOKDE5SKEUU/action/author_attestation","sign_citation":"https://pith.science/pith/R3GDJ3QQ3C2ZQDFYOKDE5SKEUU/action/citation_signature","submit_replication":"https://pith.science/pith/R3GDJ3QQ3C2ZQDFYOKDE5SKEUU/action/replication_record"}},"created_at":"2026-07-05T00:42:00.815527+00:00","updated_at":"2026-07-05T00:42:00.815527+00:00"}