{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:WXUUDIJIGMTDQF2ONCP7VIIUSB","short_pith_number":"pith:WXUUDIJI","schema_version":"1.0","canonical_sha256":"b5e941a128332638174e689ffaa114907da903d4ef718ac6135a7b25fa5ee719","source":{"kind":"arxiv","id":"2010.02824","version":2},"attestation_state":"computed","paper":{"title":"Support-set bottlenecks for video-text representation learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Alexander Hauptmann, Andrea Vedaldi, Florian Metze, Jo\\~ao Henriques, Mandela Patrick, Po-Yao Huang, Yuki Asano","submitted_at":"2020-10-06T15:38:54Z","abstract_excerpt":"The dominant paradigm for learning video-text representations -- noise contrastive learning -- increases the similarity of the representations of pairs of samples that are known to be related, such as text and video from the same sample, and pushes away the representations of all other pairs. We posit that this last behaviour is too strict, enforcing dissimilar representations even for samples that are semantically-related -- for example, visually similar videos or ones that share the same depicted action. In this paper, we propose a novel method that alleviates this by leveraging a generative"},"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":"2010.02824","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2020-10-06T15:38:54Z","cross_cats_sorted":[],"title_canon_sha256":"fd7d45119293e924f7c9377de92a402bf74329182b6f76430417bf8e6052589d","abstract_canon_sha256":"105a29a2a3254d49557b7939812e361a7b327b5dd86ac500787265732558eee3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:06:52.499389Z","signature_b64":"eenmOLKc51BwRo6QCw8mA3OUWD7e0vTcdK1frwVk/G3wZaHDsYKZFjHv6VkGUNkNwA9oA9OB9B3v/MjCLau+BA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b5e941a128332638174e689ffaa114907da903d4ef718ac6135a7b25fa5ee719","last_reissued_at":"2026-07-05T02:06:52.498953Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:06:52.498953Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Support-set bottlenecks for video-text representation learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Alexander Hauptmann, Andrea Vedaldi, Florian Metze, Jo\\~ao Henriques, Mandela Patrick, Po-Yao Huang, Yuki Asano","submitted_at":"2020-10-06T15:38:54Z","abstract_excerpt":"The dominant paradigm for learning video-text representations -- noise contrastive learning -- increases the similarity of the representations of pairs of samples that are known to be related, such as text and video from the same sample, and pushes away the representations of all other pairs. We posit that this last behaviour is too strict, enforcing dissimilar representations even for samples that are semantically-related -- for example, visually similar videos or ones that share the same depicted action. In this paper, we propose a novel method that alleviates this by leveraging a generative"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2010.02824","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/2010.02824/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":"2010.02824","created_at":"2026-07-05T02:06:52.499010+00:00"},{"alias_kind":"arxiv_version","alias_value":"2010.02824v2","created_at":"2026-07-05T02:06:52.499010+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2010.02824","created_at":"2026-07-05T02:06:52.499010+00:00"},{"alias_kind":"pith_short_12","alias_value":"WXUUDIJIGMTD","created_at":"2026-07-05T02:06:52.499010+00:00"},{"alias_kind":"pith_short_16","alias_value":"WXUUDIJIGMTDQF2O","created_at":"2026-07-05T02:06:52.499010+00:00"},{"alias_kind":"pith_short_8","alias_value":"WXUUDIJI","created_at":"2026-07-05T02:06:52.499010+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.17959","citing_title":"Text-Video Retrieval With Global-Local Contrastive Consistency Learning","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2310.01852","citing_title":"LanguageBind: Extending Video-Language Pretraining to N-modality by Language-based Semantic Alignment","ref_index":178,"is_internal_anchor":false},{"citing_arxiv_id":"2204.00598","citing_title":"Socratic Models: Composing Zero-Shot Multimodal Reasoning with Language","ref_index":92,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WXUUDIJIGMTDQF2ONCP7VIIUSB","json":"https://pith.science/pith/WXUUDIJIGMTDQF2ONCP7VIIUSB.json","graph_json":"https://pith.science/api/pith-number/WXUUDIJIGMTDQF2ONCP7VIIUSB/graph.json","events_json":"https://pith.science/api/pith-number/WXUUDIJIGMTDQF2ONCP7VIIUSB/events.json","paper":"https://pith.science/paper/WXUUDIJI"},"agent_actions":{"view_html":"https://pith.science/pith/WXUUDIJIGMTDQF2ONCP7VIIUSB","download_json":"https://pith.science/pith/WXUUDIJIGMTDQF2ONCP7VIIUSB.json","view_paper":"https://pith.science/paper/WXUUDIJI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2010.02824&json=true","fetch_graph":"https://pith.science/api/pith-number/WXUUDIJIGMTDQF2ONCP7VIIUSB/graph.json","fetch_events":"https://pith.science/api/pith-number/WXUUDIJIGMTDQF2ONCP7VIIUSB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WXUUDIJIGMTDQF2ONCP7VIIUSB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WXUUDIJIGMTDQF2ONCP7VIIUSB/action/storage_attestation","attest_author":"https://pith.science/pith/WXUUDIJIGMTDQF2ONCP7VIIUSB/action/author_attestation","sign_citation":"https://pith.science/pith/WXUUDIJIGMTDQF2ONCP7VIIUSB/action/citation_signature","submit_replication":"https://pith.science/pith/WXUUDIJIGMTDQF2ONCP7VIIUSB/action/replication_record"}},"created_at":"2026-07-05T02:06:52.499010+00:00","updated_at":"2026-07-05T02:06:52.499010+00:00"}