{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:MHSNO744LVPLACGNQ6Y25JB4ZV","short_pith_number":"pith:MHSNO744","schema_version":"1.0","canonical_sha256":"61e4d77f9c5d5eb008cd87b1aea43ccd6e512c611a71fecac4de5bab0995cae6","source":{"kind":"arxiv","id":"2007.10740","version":3},"attestation_state":"computed","paper":{"title":"Balanced Meta-Softmax for Long-Tailed Visual Recognition","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Cunjun Yu, Haiyu Zhao, Hongsheng Li, Jiawei Ren, Shuai Yi, Shunan Sheng, Xiao Ma","submitted_at":"2020-07-21T12:05:00Z","abstract_excerpt":"Deep classifiers have achieved great success in visual recognition. However, real-world data is long-tailed by nature, leading to the mismatch between training and testing distributions. In this paper, we show that the Softmax function, though used in most classification tasks, gives a biased gradient estimation under the long-tailed setup. This paper presents Balanced Softmax, an elegant unbiased extension of Softmax, to accommodate the label distribution shift between training and testing. Theoretically, we derive the generalization bound for multiclass Softmax regression and show our loss m"},"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":"2007.10740","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-07-21T12:05:00Z","cross_cats_sorted":["cs.CV","stat.ML"],"title_canon_sha256":"c0efd25bd56bad866153241287d423fd034edc99c2680794748fcd8246e075df","abstract_canon_sha256":"84e7da7521d4cf7c3564de01c7fdc1f5a648bf0d1c5fd466bcc44358de034d54"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:53:21.349727Z","signature_b64":"/W6M/0dwRTvlBPV3+pez1todMFk9trH/SDDHzf2k8SU3BCSP6dktAy+zgkYp0hNw+pxsO0G02bVl9Wnk+e3lDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"61e4d77f9c5d5eb008cd87b1aea43ccd6e512c611a71fecac4de5bab0995cae6","last_reissued_at":"2026-07-05T01:53:21.349231Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:53:21.349231Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Balanced Meta-Softmax for Long-Tailed Visual Recognition","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Cunjun Yu, Haiyu Zhao, Hongsheng Li, Jiawei Ren, Shuai Yi, Shunan Sheng, Xiao Ma","submitted_at":"2020-07-21T12:05:00Z","abstract_excerpt":"Deep classifiers have achieved great success in visual recognition. However, real-world data is long-tailed by nature, leading to the mismatch between training and testing distributions. In this paper, we show that the Softmax function, though used in most classification tasks, gives a biased gradient estimation under the long-tailed setup. This paper presents Balanced Softmax, an elegant unbiased extension of Softmax, to accommodate the label distribution shift between training and testing. Theoretically, we derive the generalization bound for multiclass Softmax regression and show our loss m"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2007.10740","kind":"arxiv","version":3},"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/2007.10740/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":"2007.10740","created_at":"2026-07-05T01:53:21.349291+00:00"},{"alias_kind":"arxiv_version","alias_value":"2007.10740v3","created_at":"2026-07-05T01:53:21.349291+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2007.10740","created_at":"2026-07-05T01:53:21.349291+00:00"},{"alias_kind":"pith_short_12","alias_value":"MHSNO744LVPL","created_at":"2026-07-05T01:53:21.349291+00:00"},{"alias_kind":"pith_short_16","alias_value":"MHSNO744LVPLACGN","created_at":"2026-07-05T01:53:21.349291+00:00"},{"alias_kind":"pith_short_8","alias_value":"MHSNO744","created_at":"2026-07-05T01:53:21.349291+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.24908","citing_title":"On the Impact of Class Imbalance on the Learning Dynamics of Deep Neural Networks:An Intuitive Insight","ref_index":31,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MHSNO744LVPLACGNQ6Y25JB4ZV","json":"https://pith.science/pith/MHSNO744LVPLACGNQ6Y25JB4ZV.json","graph_json":"https://pith.science/api/pith-number/MHSNO744LVPLACGNQ6Y25JB4ZV/graph.json","events_json":"https://pith.science/api/pith-number/MHSNO744LVPLACGNQ6Y25JB4ZV/events.json","paper":"https://pith.science/paper/MHSNO744"},"agent_actions":{"view_html":"https://pith.science/pith/MHSNO744LVPLACGNQ6Y25JB4ZV","download_json":"https://pith.science/pith/MHSNO744LVPLACGNQ6Y25JB4ZV.json","view_paper":"https://pith.science/paper/MHSNO744","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2007.10740&json=true","fetch_graph":"https://pith.science/api/pith-number/MHSNO744LVPLACGNQ6Y25JB4ZV/graph.json","fetch_events":"https://pith.science/api/pith-number/MHSNO744LVPLACGNQ6Y25JB4ZV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MHSNO744LVPLACGNQ6Y25JB4ZV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MHSNO744LVPLACGNQ6Y25JB4ZV/action/storage_attestation","attest_author":"https://pith.science/pith/MHSNO744LVPLACGNQ6Y25JB4ZV/action/author_attestation","sign_citation":"https://pith.science/pith/MHSNO744LVPLACGNQ6Y25JB4ZV/action/citation_signature","submit_replication":"https://pith.science/pith/MHSNO744LVPLACGNQ6Y25JB4ZV/action/replication_record"}},"created_at":"2026-07-05T01:53:21.349291+00:00","updated_at":"2026-07-05T01:53:21.349291+00:00"}