{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:Z4LEITEJSP34TTCZOUQJX774CI","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"37d5a4f6f6b6911af0a05b1e4044f37883fd1a23a1b84f1c93f8e54194e3ba4c","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T01:36:43Z","title_canon_sha256":"856537fdd4622b11ff2f965a6dfccd2428f52aac6093ff3c5b115f0a56176d62"},"schema_version":"1.0","source":{"id":"2605.12872","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.12872","created_at":"2026-05-18T03:09:11Z"},{"alias_kind":"arxiv_version","alias_value":"2605.12872v1","created_at":"2026-05-18T03:09:11Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.12872","created_at":"2026-05-18T03:09:11Z"},{"alias_kind":"pith_short_12","alias_value":"Z4LEITEJSP34","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"Z4LEITEJSP34TTCZ","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"Z4LEITEJ","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:b473d5560039f579520085189dee8d175b35f345cd77a325619bd26ce955db3d","target":"graph","created_at":"2026-05-18T03:09:11Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"SMA achieves strong multimodal generalization using only tens of thousands of samples. This is orders of magnitude fewer than standard approaches."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the set-based formulation with submodular mutual information captures richer cross-modal geometric structure and effectively utilizes multiple positive associations without introducing biases or requiring extensive post-hoc tuning that affects the reported gains."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"SMA uses a submodular mutual information objective on data sets to deliver competitive zero-shot classification and retrieval performance on CLIP benchmarks with only tens of thousands of samples, orders of magnitude fewer than standard approaches."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"SMA aligns images and text by optimizing submodular mutual information over sets of descriptions rather than individual pairs, enabling strong zero-shot performance with only tens of thousands of samples."}],"snapshot_sha256":"4e1704971da14415c7088e0a800e567c6c9ecf21f815d7685ecb04c04fdccea9"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"4d4512cc3b4b824f742f870a1403d8342de3aae4e79768480a484bc57141d2d4"},"paper":{"abstract_excerpt":"Despite the recent success of Multimodal Foundation Models (FMs), their reliance on massive paired datasets limits their applicability in low-data and rare-scenario settings where aligned data is scarce and expensive. A key bottleneck is the adoption of an instance-level formulation, which learns alignment by maximizing correlation between individual image-text pairs while neglecting the underlying geometric structure across modalities resulting in a modality gap across input modalities. In this paper, we propose a combinatorial paradigm for multimodal alignment that moves beyond pairwise lear","authors_text":"Anay Majee, Rishabh Iyer, Truong Pham","cross_cats":[],"headline":"SMA aligns images and text by optimizing submodular mutual information over sets of descriptions rather than individual pairs, enabling strong zero-shot performance with only tens of thousands of samples.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T01:36:43Z","title":"SMA: Submodular Modality Aligner For Data Efficient Multimodal Learning"},"references":{"count":54,"internal_anchors":5,"resolved_work":54,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Liteembed: Adapting clip to rare classes, 2026","work_id":"797ef698-5fd1-44b3-a38c-c6748e0bc479","year":2026},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Theoretical analysis of submodular information measures for targeted data subset selection.ArXiv, abs/2402.13454, 2024","work_id":"8bdfc0a7-ae56-4245-bdf7-5394f7e823f8","year":2024},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"arXiv preprint arXiv:2106.15324 , year=","work_id":"9e312208-7e3a-4663-ab37-1d9cc653c348","year":2021},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Submodularity in machine learning and artificial intelligence","work_id":"83a2d723-7834-4f2a-a012-e49eae74b5c5","year":2022},{"cited_arxiv_id":"2002.05709","doi":"","is_internal_anchor":true,"ref_index":5,"title":"A Simple Framework for Contrastive Learning of Visual Representations","work_id":"77d995ce-c44e-4692-9c54-cf8ce771464a","year":2002}],"snapshot_sha256":"e71b506350c44a0670cbbd0109066f8555609e414a443a92e123ae756e78777e"},"source":{"id":"2605.12872","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-14T20:28:21.138946Z","id":"cca2a51a-d698-480c-944c-647333612a5b","model_set":{"reader":"grok-4.3"},"one_line_summary":"SMA uses a submodular mutual information objective on data sets to deliver competitive zero-shot classification and retrieval performance on CLIP benchmarks with only tens of thousands of samples, orders of magnitude fewer than standard approaches.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"SMA aligns images and text by optimizing submodular mutual information over sets of descriptions rather than individual pairs, enabling strong zero-shot performance with only tens of thousands of samples.","strongest_claim":"SMA achieves strong multimodal generalization using only tens of thousands of samples. This is orders of magnitude fewer than standard approaches.","weakest_assumption":"That the set-based formulation with submodular mutual information captures richer cross-modal geometric structure and effectively utilizes multiple positive associations without introducing biases or requiring extensive post-hoc tuning that affects the reported gains."}},"verdict_id":"cca2a51a-d698-480c-944c-647333612a5b"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:d6e2d2fc1ed824f26dde50792e8a2f43fdf5abf7907e1071815c223d6b3a2aae","target":"record","created_at":"2026-05-18T03:09:11Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"37d5a4f6f6b6911af0a05b1e4044f37883fd1a23a1b84f1c93f8e54194e3ba4c","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T01:36:43Z","title_canon_sha256":"856537fdd4622b11ff2f965a6dfccd2428f52aac6093ff3c5b115f0a56176d62"},"schema_version":"1.0","source":{"id":"2605.12872","kind":"arxiv","version":1}},"canonical_sha256":"cf16444c8993f7c9cc5975209bfffc1237c387fc3dd03eb1bd191ee76db88ed8","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"cf16444c8993f7c9cc5975209bfffc1237c387fc3dd03eb1bd191ee76db88ed8","first_computed_at":"2026-05-18T03:09:11.309725Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T03:09:11.309725Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"QemppQEjGbDDy8AHWLidV6Iuwq7I+AiosMiaHI2Ym/vKh4UwKHOZwY1nDOly0Lia+h4oKEkc/I5J2KL/m+VDBQ==","signature_status":"signed_v1","signed_at":"2026-05-18T03:09:11.310326Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.12872","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:d6e2d2fc1ed824f26dde50792e8a2f43fdf5abf7907e1071815c223d6b3a2aae","sha256:b473d5560039f579520085189dee8d175b35f345cd77a325619bd26ce955db3d"],"state_sha256":"32b1d807d771f4ceb596c610aeb8ca69a4080535ff2171bb2676d7a319432c0f"}