{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:3VCQT6VVG4BGXCKQDVRWVYJTOD","short_pith_number":"pith:3VCQT6VV","schema_version":"1.0","canonical_sha256":"dd4509fab537026b89501d636ae13370cce349d52a0bd717c06119bd509c3a9c","source":{"kind":"arxiv","id":"2602.05845","version":2},"attestation_state":"computed","paper":{"title":"Self-Supervised Learning with a Multi-Task Latent Space Objective","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Abhishek Jha, Luc Van Gool, Marc Proesmans, Pierre-Fran\\c{c}ois De Plaen, Tinne Tuytelaars","submitted_at":"2026-02-05T16:33:30Z","abstract_excerpt":"We propose a multi-task formulation of self-predictive Siamese SSL in which each spatial transformation defines a distinct latent-space alignment task, solved by a dedicated predictor over a shared encoder. This perspective directly explains a long-standing failure of multi-crop training in self-predictive methods such as BYOL, SimSiam, and MoCo v3: a shared predictor is forced to solve heterogeneous alignment tasks simultaneously, leading to unstable optimization. Assigning one predictor per view type resolves this interference, unlocking linear evaluation gains of 3.8-4\\% across frameworks. "},"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":"2602.05845","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-02-05T16:33:30Z","cross_cats_sorted":[],"title_canon_sha256":"5bb0c579236cd3a92475ec0edd160ac36cc55dd23f18fc159b3744f158060915","abstract_canon_sha256":"61e38843648d1b5adf7dde8b17d27f05e979efdf59fe9f265697cabd2f339ca4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T02:07:19.582825Z","signature_b64":"yMK8mJEuNLag1ujqOiqpUST+H4gbVzK5K7H+8dX1IR+jBMzn6/u3IDodQgpCAYZgCfxrQjp3tdTYXsHGGMhZCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"dd4509fab537026b89501d636ae13370cce349d52a0bd717c06119bd509c3a9c","last_reissued_at":"2026-06-09T02:07:19.582169Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T02:07:19.582169Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Self-Supervised Learning with a Multi-Task Latent Space Objective","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Abhishek Jha, Luc Van Gool, Marc Proesmans, Pierre-Fran\\c{c}ois De Plaen, Tinne Tuytelaars","submitted_at":"2026-02-05T16:33:30Z","abstract_excerpt":"We propose a multi-task formulation of self-predictive Siamese SSL in which each spatial transformation defines a distinct latent-space alignment task, solved by a dedicated predictor over a shared encoder. This perspective directly explains a long-standing failure of multi-crop training in self-predictive methods such as BYOL, SimSiam, and MoCo v3: a shared predictor is forced to solve heterogeneous alignment tasks simultaneously, leading to unstable optimization. Assigning one predictor per view type resolves this interference, unlocking linear evaluation gains of 3.8-4\\% across frameworks. "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.05845","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/2602.05845/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":"2602.05845","created_at":"2026-06-09T02:07:19.582232+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.05845v2","created_at":"2026-06-09T02:07:19.582232+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.05845","created_at":"2026-06-09T02:07:19.582232+00:00"},{"alias_kind":"pith_short_12","alias_value":"3VCQT6VVG4BG","created_at":"2026-06-09T02:07:19.582232+00:00"},{"alias_kind":"pith_short_16","alias_value":"3VCQT6VVG4BGXCKQ","created_at":"2026-06-09T02:07:19.582232+00:00"},{"alias_kind":"pith_short_8","alias_value":"3VCQT6VV","created_at":"2026-06-09T02:07:19.582232+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/3VCQT6VVG4BGXCKQDVRWVYJTOD","json":"https://pith.science/pith/3VCQT6VVG4BGXCKQDVRWVYJTOD.json","graph_json":"https://pith.science/api/pith-number/3VCQT6VVG4BGXCKQDVRWVYJTOD/graph.json","events_json":"https://pith.science/api/pith-number/3VCQT6VVG4BGXCKQDVRWVYJTOD/events.json","paper":"https://pith.science/paper/3VCQT6VV"},"agent_actions":{"view_html":"https://pith.science/pith/3VCQT6VVG4BGXCKQDVRWVYJTOD","download_json":"https://pith.science/pith/3VCQT6VVG4BGXCKQDVRWVYJTOD.json","view_paper":"https://pith.science/paper/3VCQT6VV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.05845&json=true","fetch_graph":"https://pith.science/api/pith-number/3VCQT6VVG4BGXCKQDVRWVYJTOD/graph.json","fetch_events":"https://pith.science/api/pith-number/3VCQT6VVG4BGXCKQDVRWVYJTOD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3VCQT6VVG4BGXCKQDVRWVYJTOD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3VCQT6VVG4BGXCKQDVRWVYJTOD/action/storage_attestation","attest_author":"https://pith.science/pith/3VCQT6VVG4BGXCKQDVRWVYJTOD/action/author_attestation","sign_citation":"https://pith.science/pith/3VCQT6VVG4BGXCKQDVRWVYJTOD/action/citation_signature","submit_replication":"https://pith.science/pith/3VCQT6VVG4BGXCKQDVRWVYJTOD/action/replication_record"}},"created_at":"2026-06-09T02:07:19.582232+00:00","updated_at":"2026-06-09T02:07:19.582232+00:00"}