{"paper":{"title":"Accelerated Sequential Flow Matching: A Bayesian Filtering Perspective","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Sequential Bayesian Flow Matching reuses the previous posterior as a source distribution to accelerate sampling from streaming observations.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bo Dai, Hans Hao-Hsun Hsu, Junran Wang, Pan Li, Yinan Huang","submitted_at":"2026-02-05T05:37:14Z","abstract_excerpt":"Sequential probabilistic inference from streaming observations requires modeling distributions over future trajectories as new observations arrive. Although diffusion and flow-matching models are effective at capturing high-dimensional, multimodal distributions, their deployment in real-time streaming settings typically relies on repeatedly sampling from a non-informative initial distribution. This results in substantial inference latency, particularly when multiple samples are needed to characterize the predictive distribution. In this work, we introduce Sequential Bayesian Flow Matching, a f"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"By using the previous belief as an informative source distribution, it enables substantially faster sampling than naive resampling from scratch while achieving performance competitive with full-step diffusion on distributional metrics.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That a learned flow can reliably transport the full posterior distribution (including multimodality) from one time step to the next without accumulating approximation error or requiring retraining when the observation model changes.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Sequential Bayesian Flow Matching accelerates flow-based sampling for streaming probabilistic inference by transporting posteriors recursively like Bayesian filters rather than restarting from noise each step.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Sequential Bayesian Flow Matching reuses the previous posterior as a source distribution to accelerate sampling from streaming observations.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"53da81f087669c8aae81a6b6547a851b1f4c67752db5974c61163d6af138f386"},"source":{"id":"2602.05319","kind":"arxiv","version":3},"verdict":{"id":"4a48b38d-2752-4db3-b5e7-fc56265d0e48","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T07:28:04.950920Z","strongest_claim":"By using the previous belief as an informative source distribution, it enables substantially faster sampling than naive resampling from scratch while achieving performance competitive with full-step diffusion on distributional metrics.","one_line_summary":"Sequential Bayesian Flow Matching accelerates flow-based sampling for streaming probabilistic inference by transporting posteriors recursively like Bayesian filters rather than restarting from noise each step.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That a learned flow can reliably transport the full posterior distribution (including multimodality) from one time step to the next without accumulating approximation error or requiring retraining when the observation model changes.","pith_extraction_headline":"Sequential Bayesian Flow Matching reuses the previous posterior as a source distribution to accelerate sampling from streaming observations."},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"27e09ef0999223a651f718af491f5ee6070a22877db8e628c758b516da5c7352"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}