{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:FNKHUXSW2ENG54D7T4H57BZSSO","short_pith_number":"pith:FNKHUXSW","schema_version":"1.0","canonical_sha256":"2b547a5e56d11a6ef07f9f0fdf8732938aae307982ee83e856035505e8514942","source":{"kind":"arxiv","id":"2605.15622","version":2},"attestation_state":"computed","paper":{"title":"Position: Zeroth-Order Optimization in Deep Learning Is Underexplored, Not Underpowered","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bhavya Kailkhura, Changsheng Wang, Chongyu Fan, James Diffenderfer, Sijia Liu, Soumyadeep Pal, Yancheng Huang, Yicheng Lang, Yihua Zhang","submitted_at":"2026-05-15T05:11:43Z","abstract_excerpt":"Zeroth-order (ZO) optimization, learning from finite differences of function evaluations without backpropagation, has recently regained attention in deep learning due to its memory efficiency and applicability to gray- or black-box pipelines. Yet, ZO methods are often dismissed as fundamentally unscalable because of estimator variance and unfavorable query complexity. We argue that this conclusion might be misguided: ZO optimization is underexplored, not underpowered. We show that many perceived limitations stem from myopic development practices, most notably full-space, element-wise, estimato"},"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":"2605.15622","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-15T05:11:43Z","cross_cats_sorted":[],"title_canon_sha256":"b94c4e5818827ce977fd2ac9205a7d2e141abe39656aacaa08cf38bd8ebfc264","abstract_canon_sha256":"e8cd4034e8fe7eaf1b22c68941a3d5c8f8163abad8e3a2a5e99f668a6ae00ffe"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:04:38.194179Z","signature_b64":"YpuiGSchpOltjXWMNcW4MF9TR8/g10tnmhqyj4JiXw6KV4dyPhFuDsfKqILza65w3pRq+ZizfwWda9J7Y2XQAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2b547a5e56d11a6ef07f9f0fdf8732938aae307982ee83e856035505e8514942","last_reissued_at":"2026-05-20T00:04:38.193260Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:04:38.193260Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Position: Zeroth-Order Optimization in Deep Learning Is Underexplored, Not Underpowered","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bhavya Kailkhura, Changsheng Wang, Chongyu Fan, James Diffenderfer, Sijia Liu, Soumyadeep Pal, Yancheng Huang, Yicheng Lang, Yihua Zhang","submitted_at":"2026-05-15T05:11:43Z","abstract_excerpt":"Zeroth-order (ZO) optimization, learning from finite differences of function evaluations without backpropagation, has recently regained attention in deep learning due to its memory efficiency and applicability to gray- or black-box pipelines. Yet, ZO methods are often dismissed as fundamentally unscalable because of estimator variance and unfavorable query complexity. We argue that this conclusion might be misguided: ZO optimization is underexplored, not underpowered. We show that many perceived limitations stem from myopic development practices, most notably full-space, element-wise, estimato"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.15622","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/2605.15622/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":"2605.15622","created_at":"2026-05-20T00:04:38.193434+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.15622v2","created_at":"2026-05-20T00:04:38.193434+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.15622","created_at":"2026-05-20T00:04:38.193434+00:00"},{"alias_kind":"pith_short_12","alias_value":"FNKHUXSW2ENG","created_at":"2026-05-20T00:04:38.193434+00:00"},{"alias_kind":"pith_short_16","alias_value":"FNKHUXSW2ENG54D7","created_at":"2026-05-20T00:04:38.193434+00:00"},{"alias_kind":"pith_short_8","alias_value":"FNKHUXSW","created_at":"2026-05-20T00:04:38.193434+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FNKHUXSW2ENG54D7T4H57BZSSO","json":"https://pith.science/pith/FNKHUXSW2ENG54D7T4H57BZSSO.json","graph_json":"https://pith.science/api/pith-number/FNKHUXSW2ENG54D7T4H57BZSSO/graph.json","events_json":"https://pith.science/api/pith-number/FNKHUXSW2ENG54D7T4H57BZSSO/events.json","paper":"https://pith.science/paper/FNKHUXSW"},"agent_actions":{"view_html":"https://pith.science/pith/FNKHUXSW2ENG54D7T4H57BZSSO","download_json":"https://pith.science/pith/FNKHUXSW2ENG54D7T4H57BZSSO.json","view_paper":"https://pith.science/paper/FNKHUXSW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.15622&json=true","fetch_graph":"https://pith.science/api/pith-number/FNKHUXSW2ENG54D7T4H57BZSSO/graph.json","fetch_events":"https://pith.science/api/pith-number/FNKHUXSW2ENG54D7T4H57BZSSO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FNKHUXSW2ENG54D7T4H57BZSSO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FNKHUXSW2ENG54D7T4H57BZSSO/action/storage_attestation","attest_author":"https://pith.science/pith/FNKHUXSW2ENG54D7T4H57BZSSO/action/author_attestation","sign_citation":"https://pith.science/pith/FNKHUXSW2ENG54D7T4H57BZSSO/action/citation_signature","submit_replication":"https://pith.science/pith/FNKHUXSW2ENG54D7T4H57BZSSO/action/replication_record"}},"created_at":"2026-05-20T00:04:38.193434+00:00","updated_at":"2026-05-20T00:04:38.193434+00:00"}