{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:UPASTQJTLQAISYWFHA4HZX6PJE","short_pith_number":"pith:UPASTQJT","schema_version":"1.0","canonical_sha256":"a3c129c1335c008962c538387cdfcf492ff763b3e827821757f7fece6d639801","source":{"kind":"arxiv","id":"2305.17326","version":7},"attestation_state":"computed","paper":{"title":"Matrix Information Theory for Self-Supervised Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.LG","authors_text":"Jingqin Yang, Weiran Huang, Yang Yuan, Yifan Zhang, Zhiquan Tan","submitted_at":"2023-05-27T02:04:25Z","abstract_excerpt":"The maximum entropy encoding framework provides a unified perspective for many non-contrastive learning methods like SimSiam, Barlow Twins, and MEC. Inspired by this framework, we introduce Matrix-SSL, a novel approach that leverages matrix information theory to interpret the maximum entropy encoding loss as matrix uniformity loss. Furthermore, Matrix-SSL enhances the maximum entropy encoding method by seamlessly incorporating matrix alignment loss, directly aligning covariance matrices in different branches. Experimental results reveal that Matrix-SSL outperforms state-of-the-art methods on t"},"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":"2305.17326","kind":"arxiv","version":7},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-05-27T02:04:25Z","cross_cats_sorted":["cs.AI","cs.CV"],"title_canon_sha256":"811cd6b47bf928143739d875a704b531cae6a7c9528dfcb87744c2ac395b1a75","abstract_canon_sha256":"fd950bf898979c72ac6963c569d624af0b88d3a38101d69852feca4698276b63"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:06:53.833338Z","signature_b64":"Oq09Hr80r7ph42AM/R9Mpfb4lQ3x1aL+nVcOAHWKb5Z4BsJWrLb/IKzrJZvbyRMUvAzKHzb15LC+01U+0hkWCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a3c129c1335c008962c538387cdfcf492ff763b3e827821757f7fece6d639801","last_reissued_at":"2026-07-05T09:06:53.832862Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:06:53.832862Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Matrix Information Theory for Self-Supervised Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.LG","authors_text":"Jingqin Yang, Weiran Huang, Yang Yuan, Yifan Zhang, Zhiquan Tan","submitted_at":"2023-05-27T02:04:25Z","abstract_excerpt":"The maximum entropy encoding framework provides a unified perspective for many non-contrastive learning methods like SimSiam, Barlow Twins, and MEC. Inspired by this framework, we introduce Matrix-SSL, a novel approach that leverages matrix information theory to interpret the maximum entropy encoding loss as matrix uniformity loss. Furthermore, Matrix-SSL enhances the maximum entropy encoding method by seamlessly incorporating matrix alignment loss, directly aligning covariance matrices in different branches. Experimental results reveal that Matrix-SSL outperforms state-of-the-art methods on t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.17326","kind":"arxiv","version":7},"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/2305.17326/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":"2305.17326","created_at":"2026-07-05T09:06:53.832916+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.17326v7","created_at":"2026-07-05T09:06:53.832916+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.17326","created_at":"2026-07-05T09:06:53.832916+00:00"},{"alias_kind":"pith_short_12","alias_value":"UPASTQJTLQAI","created_at":"2026-07-05T09:06:53.832916+00:00"},{"alias_kind":"pith_short_16","alias_value":"UPASTQJTLQAISYWF","created_at":"2026-07-05T09:06:53.832916+00:00"},{"alias_kind":"pith_short_8","alias_value":"UPASTQJT","created_at":"2026-07-05T09:06:53.832916+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.22873","citing_title":"SingGuard: A Policy-Adaptive Multimodal LLM Guardrail with Dynamic Reasoning","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2606.22873","citing_title":"SingGuard: A Policy-Adaptive Multimodal LLM Guardrail with Dynamic Reasoning","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2605.29720","citing_title":"Efficient, Validation-Free Intrinsic Quality Estimation for Large-Scale Face Recognition Datasets","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2511.08544","citing_title":"LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristics","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01967","citing_title":"MER-DG: Modality-Entropy Regularization for Multimodal Domain Generalization","ref_index":24,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UPASTQJTLQAISYWFHA4HZX6PJE","json":"https://pith.science/pith/UPASTQJTLQAISYWFHA4HZX6PJE.json","graph_json":"https://pith.science/api/pith-number/UPASTQJTLQAISYWFHA4HZX6PJE/graph.json","events_json":"https://pith.science/api/pith-number/UPASTQJTLQAISYWFHA4HZX6PJE/events.json","paper":"https://pith.science/paper/UPASTQJT"},"agent_actions":{"view_html":"https://pith.science/pith/UPASTQJTLQAISYWFHA4HZX6PJE","download_json":"https://pith.science/pith/UPASTQJTLQAISYWFHA4HZX6PJE.json","view_paper":"https://pith.science/paper/UPASTQJT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.17326&json=true","fetch_graph":"https://pith.science/api/pith-number/UPASTQJTLQAISYWFHA4HZX6PJE/graph.json","fetch_events":"https://pith.science/api/pith-number/UPASTQJTLQAISYWFHA4HZX6PJE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UPASTQJTLQAISYWFHA4HZX6PJE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UPASTQJTLQAISYWFHA4HZX6PJE/action/storage_attestation","attest_author":"https://pith.science/pith/UPASTQJTLQAISYWFHA4HZX6PJE/action/author_attestation","sign_citation":"https://pith.science/pith/UPASTQJTLQAISYWFHA4HZX6PJE/action/citation_signature","submit_replication":"https://pith.science/pith/UPASTQJTLQAISYWFHA4HZX6PJE/action/replication_record"}},"created_at":"2026-07-05T09:06:53.832916+00:00","updated_at":"2026-07-05T09:06:53.832916+00:00"}