{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:OP6JBWTFSXBOYUSP4ATGQYMSPI","short_pith_number":"pith:OP6JBWTF","schema_version":"1.0","canonical_sha256":"73fc90da6595c2ec524fe0266861927a03c8f50ff1c745b8bad5430137ab6530","source":{"kind":"arxiv","id":"2604.04958","version":3},"attestation_state":"computed","paper":{"title":"CalM: A Self-Supervised Foundation Model for Population Dynamics in Calcium Imaging Data","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A self-supervised model pretrained on calcium traces forecasts neural population dynamics better than specialized baselines and adapts to decode behavior.","cross_cats":["cs.AI","q-bio.NC"],"primary_cat":"q-bio.QM","authors_text":"Qichen Qian, Xinhong Xu, Yimeng Zhang, Yuanlong Zhang","submitted_at":"2026-04-03T13:46:41Z","abstract_excerpt":"Recent work suggests that large-scale, multi-animal modeling can significantly improve neural recording analysis. However, for functional calcium traces, existing approaches remain task-specific, limiting transfer across common neuroscience objectives. To address this challenge, we propose \\textbf{CalM}, a self-supervised neural foundation model trained solely on neuronal calcium traces and adaptable to multiple downstream tasks, including forecasting and decoding. Our key contribution is a pretraining framework, composed of a high-performance tokenizer mapping single-neuron traces into a shar"},"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":true},"canonical_record":{"source":{"id":"2604.04958","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"q-bio.QM","submitted_at":"2026-04-03T13:46:41Z","cross_cats_sorted":["cs.AI","q-bio.NC"],"title_canon_sha256":"743659becab8043dcc2f8cd9227d510a4af00ac3f85dcc1443849ac5d9de3a35","abstract_canon_sha256":"10cc471c22c13dc4e5aac9c2166929d863946cd7202548701c030ee454216853"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T02:04:52.785333Z","signature_b64":"IdDMBk1wFdDvlfrvBBT7zFBBEEz7Q0GGFRg+sH5il++tEDx1Y9Hvg92iXjcv0smLJ1nK1i8ZxT4bPkJ/E/E5BA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"73fc90da6595c2ec524fe0266861927a03c8f50ff1c745b8bad5430137ab6530","last_reissued_at":"2026-06-02T02:04:52.784866Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T02:04:52.784866Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CalM: A Self-Supervised Foundation Model for Population Dynamics in Calcium Imaging Data","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A self-supervised model pretrained on calcium traces forecasts neural population dynamics better than specialized baselines and adapts to decode behavior.","cross_cats":["cs.AI","q-bio.NC"],"primary_cat":"q-bio.QM","authors_text":"Qichen Qian, Xinhong Xu, Yimeng Zhang, Yuanlong Zhang","submitted_at":"2026-04-03T13:46:41Z","abstract_excerpt":"Recent work suggests that large-scale, multi-animal modeling can significantly improve neural recording analysis. However, for functional calcium traces, existing approaches remain task-specific, limiting transfer across common neuroscience objectives. To address this challenge, we propose \\textbf{CalM}, a self-supervised neural foundation model trained solely on neuronal calcium traces and adaptable to multiple downstream tasks, including forecasting and decoding. Our key contribution is a pretraining framework, composed of a high-performance tokenizer mapping single-neuron traces into a shar"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"On the neural population dynamics forecasting task, CalM outperforms strong specialized baselines after pretraining. With a task-specific head, CalM further adapts to the behavior decoding task and achieves superior results compared with supervised decoding models.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the self-supervised pretraining framework with the proposed tokenizer and dual-axis transformer learns representations that transfer effectively to multiple downstream tasks without requiring extensive task-specific architectural changes or data curation.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"CalM uses a discrete tokenizer and dual-axis autoregressive transformer pretrained self-supervised on calcium traces to outperform specialized baselines on population dynamics forecasting and adapt to superior behavior decoding.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A self-supervised model pretrained on calcium traces forecasts neural population dynamics better than specialized baselines and adapts to decode behavior.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"06b0111031f698f7bd69adba667445ec8c5f048474b14d993d16fc781d8935e6"},"source":{"id":"2604.04958","kind":"arxiv","version":3},"verdict":{"id":"829685fb-b9df-4dfa-9224-1e0896e73a38","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-13T18:27:10.581242Z","strongest_claim":"On the neural population dynamics forecasting task, CalM outperforms strong specialized baselines after pretraining. With a task-specific head, CalM further adapts to the behavior decoding task and achieves superior results compared with supervised decoding models.","one_line_summary":"CalM uses a discrete tokenizer and dual-axis autoregressive transformer pretrained self-supervised on calcium traces to outperform specialized baselines on population dynamics forecasting and adapt to superior behavior decoding.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the self-supervised pretraining framework with the proposed tokenizer and dual-axis transformer learns representations that transfer effectively to multiple downstream tasks without requiring extensive task-specific architectural changes or data curation.","pith_extraction_headline":"A self-supervised model pretrained on calcium traces forecasts neural population dynamics better than specialized baselines and adapts to decode behavior."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.04958/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":2,"snapshot_sha256":"851f205b46f4a06c91b270761801e31eb21f29f9417e12a50029688288b3be6c"},"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":"2604.04958","created_at":"2026-06-02T02:04:52.784930+00:00"},{"alias_kind":"arxiv_version","alias_value":"2604.04958v3","created_at":"2026-06-02T02:04:52.784930+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.04958","created_at":"2026-06-02T02:04:52.784930+00:00"},{"alias_kind":"pith_short_12","alias_value":"OP6JBWTFSXBO","created_at":"2026-06-02T02:04:52.784930+00:00"},{"alias_kind":"pith_short_16","alias_value":"OP6JBWTFSXBOYUSP","created_at":"2026-06-02T02:04:52.784930+00:00"},{"alias_kind":"pith_short_8","alias_value":"OP6JBWTF","created_at":"2026-06-02T02:04:52.784930+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OP6JBWTFSXBOYUSP4ATGQYMSPI","json":"https://pith.science/pith/OP6JBWTFSXBOYUSP4ATGQYMSPI.json","graph_json":"https://pith.science/api/pith-number/OP6JBWTFSXBOYUSP4ATGQYMSPI/graph.json","events_json":"https://pith.science/api/pith-number/OP6JBWTFSXBOYUSP4ATGQYMSPI/events.json","paper":"https://pith.science/paper/OP6JBWTF"},"agent_actions":{"view_html":"https://pith.science/pith/OP6JBWTFSXBOYUSP4ATGQYMSPI","download_json":"https://pith.science/pith/OP6JBWTFSXBOYUSP4ATGQYMSPI.json","view_paper":"https://pith.science/paper/OP6JBWTF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2604.04958&json=true","fetch_graph":"https://pith.science/api/pith-number/OP6JBWTFSXBOYUSP4ATGQYMSPI/graph.json","fetch_events":"https://pith.science/api/pith-number/OP6JBWTFSXBOYUSP4ATGQYMSPI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OP6JBWTFSXBOYUSP4ATGQYMSPI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OP6JBWTFSXBOYUSP4ATGQYMSPI/action/storage_attestation","attest_author":"https://pith.science/pith/OP6JBWTFSXBOYUSP4ATGQYMSPI/action/author_attestation","sign_citation":"https://pith.science/pith/OP6JBWTFSXBOYUSP4ATGQYMSPI/action/citation_signature","submit_replication":"https://pith.science/pith/OP6JBWTFSXBOYUSP4ATGQYMSPI/action/replication_record"}},"created_at":"2026-06-02T02:04:52.784930+00:00","updated_at":"2026-06-02T02:04:52.784930+00:00"}