{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:O7GCJ6XULIYWDY74FJTYZZMIWS","short_pith_number":"pith:O7GCJ6XU","schema_version":"1.0","canonical_sha256":"77cc24faf45a3161e3fc2a678ce588b499d3d0d0fe199ae6209e67dfd49cbb8a","source":{"kind":"arxiv","id":"2203.12221","version":1},"attestation_state":"computed","paper":{"title":"Modality Competition: What Makes Joint Training of Multi-modal Network Fail in Deep Learning? (Provably)","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chang Zhou, Hongxia Yang, Junyang Lin, Longbo Huang, Yu Huang","submitted_at":"2022-03-23T06:21:53Z","abstract_excerpt":"Despite the remarkable success of deep multi-modal learning in practice, it has not been well-explained in theory. Recently, it has been observed that the best uni-modal network outperforms the jointly trained multi-modal network, which is counter-intuitive since multiple signals generally bring more information. This work provides a theoretical explanation for the emergence of such performance gap in neural networks for the prevalent joint training framework. Based on a simplified data distribution that captures the realistic property of multi-modal data, we prove that for the multi-modal lat"},"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":"2203.12221","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-03-23T06:21:53Z","cross_cats_sorted":[],"title_canon_sha256":"e68e9501b38768fabdb921699f268a027e696c4c716087e25b62b5fc3792878b","abstract_canon_sha256":"e79c6061f94310aaae63dde256eda2ca483c3e89a145fc2abebd61105d0806b7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:07:57.068016Z","signature_b64":"X20dgwffH8HHvOpMM3y1zwT4fP64Tfa2jA6GtOvwZVCwVFF/EltfGUDpkDK4ozbxYAn4ur0cLntGB6iqwERpCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"77cc24faf45a3161e3fc2a678ce588b499d3d0d0fe199ae6209e67dfd49cbb8a","last_reissued_at":"2026-07-05T04:07:57.067595Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:07:57.067595Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Modality Competition: What Makes Joint Training of Multi-modal Network Fail in Deep Learning? (Provably)","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chang Zhou, Hongxia Yang, Junyang Lin, Longbo Huang, Yu Huang","submitted_at":"2022-03-23T06:21:53Z","abstract_excerpt":"Despite the remarkable success of deep multi-modal learning in practice, it has not been well-explained in theory. Recently, it has been observed that the best uni-modal network outperforms the jointly trained multi-modal network, which is counter-intuitive since multiple signals generally bring more information. This work provides a theoretical explanation for the emergence of such performance gap in neural networks for the prevalent joint training framework. Based on a simplified data distribution that captures the realistic property of multi-modal data, we prove that for the multi-modal lat"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2203.12221","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":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2203.12221/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":"2203.12221","created_at":"2026-07-05T04:07:57.067656+00:00"},{"alias_kind":"arxiv_version","alias_value":"2203.12221v1","created_at":"2026-07-05T04:07:57.067656+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2203.12221","created_at":"2026-07-05T04:07:57.067656+00:00"},{"alias_kind":"pith_short_12","alias_value":"O7GCJ6XULIYW","created_at":"2026-07-05T04:07:57.067656+00:00"},{"alias_kind":"pith_short_16","alias_value":"O7GCJ6XULIYWDY74","created_at":"2026-07-05T04:07:57.067656+00:00"},{"alias_kind":"pith_short_8","alias_value":"O7GCJ6XU","created_at":"2026-07-05T04:07:57.067656+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.11614","citing_title":"Information-Theoretic Decomposition for Multimodal Interaction Learning","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01856","citing_title":"Boosting Multimodal Federated Learning via Chained Modality Optimization","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2602.10207","citing_title":"Optimizing Deep Learning Photometric Redshifts for the Roman Space Telescope with HST/CANDELS","ref_index":85,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/O7GCJ6XULIYWDY74FJTYZZMIWS","json":"https://pith.science/pith/O7GCJ6XULIYWDY74FJTYZZMIWS.json","graph_json":"https://pith.science/api/pith-number/O7GCJ6XULIYWDY74FJTYZZMIWS/graph.json","events_json":"https://pith.science/api/pith-number/O7GCJ6XULIYWDY74FJTYZZMIWS/events.json","paper":"https://pith.science/paper/O7GCJ6XU"},"agent_actions":{"view_html":"https://pith.science/pith/O7GCJ6XULIYWDY74FJTYZZMIWS","download_json":"https://pith.science/pith/O7GCJ6XULIYWDY74FJTYZZMIWS.json","view_paper":"https://pith.science/paper/O7GCJ6XU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2203.12221&json=true","fetch_graph":"https://pith.science/api/pith-number/O7GCJ6XULIYWDY74FJTYZZMIWS/graph.json","fetch_events":"https://pith.science/api/pith-number/O7GCJ6XULIYWDY74FJTYZZMIWS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/O7GCJ6XULIYWDY74FJTYZZMIWS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/O7GCJ6XULIYWDY74FJTYZZMIWS/action/storage_attestation","attest_author":"https://pith.science/pith/O7GCJ6XULIYWDY74FJTYZZMIWS/action/author_attestation","sign_citation":"https://pith.science/pith/O7GCJ6XULIYWDY74FJTYZZMIWS/action/citation_signature","submit_replication":"https://pith.science/pith/O7GCJ6XULIYWDY74FJTYZZMIWS/action/replication_record"}},"created_at":"2026-07-05T04:07:57.067656+00:00","updated_at":"2026-07-05T04:07:57.067656+00:00"}