{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:S56DRMWINSZKJXRZ4SJXQAX52Y","short_pith_number":"pith:S56DRMWI","schema_version":"1.0","canonical_sha256":"977c38b2c86cb2a4de39e4937802fdd60cd9e2751298cb63080c088a6d518e9d","source":{"kind":"arxiv","id":"2412.05244","version":1},"attestation_state":"computed","paper":{"title":"Enhancing Foundation Models for Time Series Forecasting via Wavelet-based Tokenization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Abdul Fatir Ansari, Andrew Gordon Wilson, Boran Han, Christos Faloutsos, Danielle C. Maddix, Luca Masserano, Michael W. Mahoney, Syama Rangapuram, Xiyuan Zhang, Youngsuk Park, Yuyang Wang","submitted_at":"2024-12-06T18:22:59Z","abstract_excerpt":"How to best develop foundational models for time series forecasting remains an important open question. Tokenization is a crucial consideration in this effort: what is an effective discrete vocabulary for a real-valued sequential input? To address this question, we develop WaveToken, a wavelet-based tokenizer that allows models to learn complex representations directly in the space of time-localized frequencies. Our method first scales and decomposes the input time series, then thresholds and quantizes the wavelet coefficients, and finally pre-trains an autoregressive model to forecast coeffic"},"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":"2412.05244","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-12-06T18:22:59Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"4b0205bfe2f8833d52746c3655414951b5dd9cefcaf56bb8bfe7ec538ffce06e","abstract_canon_sha256":"8508eb1a2fe86e31609175c7759f40814f1297e37cc73d0bcde66259e568b8e3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:45:30.772542Z","signature_b64":"qpU/m4zo0wdlGqsAADWcWDC0Zmst4CNoji7RDzNN57khioVJSlm1VlxM8KlxLWZ5r5RiSSf5dForEiWuIO9EAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"977c38b2c86cb2a4de39e4937802fdd60cd9e2751298cb63080c088a6d518e9d","last_reissued_at":"2026-07-05T09:45:30.771987Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:45:30.771987Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Enhancing Foundation Models for Time Series Forecasting via Wavelet-based Tokenization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Abdul Fatir Ansari, Andrew Gordon Wilson, Boran Han, Christos Faloutsos, Danielle C. Maddix, Luca Masserano, Michael W. Mahoney, Syama Rangapuram, Xiyuan Zhang, Youngsuk Park, Yuyang Wang","submitted_at":"2024-12-06T18:22:59Z","abstract_excerpt":"How to best develop foundational models for time series forecasting remains an important open question. Tokenization is a crucial consideration in this effort: what is an effective discrete vocabulary for a real-valued sequential input? To address this question, we develop WaveToken, a wavelet-based tokenizer that allows models to learn complex representations directly in the space of time-localized frequencies. Our method first scales and decomposes the input time series, then thresholds and quantizes the wavelet coefficients, and finally pre-trains an autoregressive model to forecast coeffic"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.05244","kind":"arxiv","version":1},"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/2412.05244/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":"2412.05244","created_at":"2026-07-05T09:45:30.772052+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.05244v1","created_at":"2026-07-05T09:45:30.772052+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.05244","created_at":"2026-07-05T09:45:30.772052+00:00"},{"alias_kind":"pith_short_12","alias_value":"S56DRMWINSZK","created_at":"2026-07-05T09:45:30.772052+00:00"},{"alias_kind":"pith_short_16","alias_value":"S56DRMWINSZKJXRZ","created_at":"2026-07-05T09:45:30.772052+00:00"},{"alias_kind":"pith_short_8","alias_value":"S56DRMWI","created_at":"2026-07-05T09:45:30.772052+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.02912","citing_title":"Data-Driven Forecasting of three-Component Seismograms Using Transformer Architectures","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12026","citing_title":"Spectral Vision Transformer for Efficient Tokenization with Limited Data","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08289","citing_title":"What If We Let Forecasting Forget? A Sparse Bottleneck for Cross-Variable Dependencies","ref_index":66,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/S56DRMWINSZKJXRZ4SJXQAX52Y","json":"https://pith.science/pith/S56DRMWINSZKJXRZ4SJXQAX52Y.json","graph_json":"https://pith.science/api/pith-number/S56DRMWINSZKJXRZ4SJXQAX52Y/graph.json","events_json":"https://pith.science/api/pith-number/S56DRMWINSZKJXRZ4SJXQAX52Y/events.json","paper":"https://pith.science/paper/S56DRMWI"},"agent_actions":{"view_html":"https://pith.science/pith/S56DRMWINSZKJXRZ4SJXQAX52Y","download_json":"https://pith.science/pith/S56DRMWINSZKJXRZ4SJXQAX52Y.json","view_paper":"https://pith.science/paper/S56DRMWI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.05244&json=true","fetch_graph":"https://pith.science/api/pith-number/S56DRMWINSZKJXRZ4SJXQAX52Y/graph.json","fetch_events":"https://pith.science/api/pith-number/S56DRMWINSZKJXRZ4SJXQAX52Y/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/S56DRMWINSZKJXRZ4SJXQAX52Y/action/timestamp_anchor","attest_storage":"https://pith.science/pith/S56DRMWINSZKJXRZ4SJXQAX52Y/action/storage_attestation","attest_author":"https://pith.science/pith/S56DRMWINSZKJXRZ4SJXQAX52Y/action/author_attestation","sign_citation":"https://pith.science/pith/S56DRMWINSZKJXRZ4SJXQAX52Y/action/citation_signature","submit_replication":"https://pith.science/pith/S56DRMWINSZKJXRZ4SJXQAX52Y/action/replication_record"}},"created_at":"2026-07-05T09:45:30.772052+00:00","updated_at":"2026-07-05T09:45:30.772052+00:00"}