{"paper":{"title":"SceneSelect: Selective Learning for Trajectory Scene Classification and Expert Scheduling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"SceneSelect classifies input scenes from geometry and motion features then routes each trajectory to the best expert predictor.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Deshun Xia, Weijie Zhu, Xinrun Wang, Yuxi Sun","submitted_at":"2026-04-27T14:15:44Z","abstract_excerpt":"Accurate trajectory prediction is fundamentally challenging due to high scene heterogeneity - the severe variance in motion velocity, spatial density, and interaction patterns across different real-world environments. However, most existing approaches typically train a single unified model, expecting a fixed-capacity architecture to generalize universally across all possible scenarios. This conventional model-centric paradigm is fundamentally flawed when confronting such extreme heterogeneity, inevitably leading to a severe generalization gap, degraded accuracy, and massive computational waste"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Extensive experiments on three public benchmarks (ETH-UCY, SDD, and NBA) demonstrate that our method consistently outperforms strong single-model and ensemble baselines, achieving an average improvement of 10.5%, showcasing the effectiveness of scene-aware selective learning.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That unsupervised clustering on geometric and kinematic features produces a stable, generalizable latent scene taxonomy, and that the decoupled classification and scheduling modules enable seamless integration with off-the-shelf models and new datasets without joint retraining or performance loss.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"SceneSelect discovers a latent scene taxonomy through clustering, trains a decoupled classifier to assign inputs, and uses a scheduling policy to dispatch to optimal expert trajectory predictors, reporting 10.5% average gains over baselines on ETH-UCY, SDD, and NBA.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"SceneSelect classifies input scenes from geometry and motion features then routes each trajectory to the best expert predictor.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"7c03aaaa25413fa610215bbfbb7ac65ba1d484a4a70d5e051c8499777e765453"},"source":{"id":"2604.24514","kind":"arxiv","version":2},"verdict":{"id":"e626397f-9bfe-455c-92c5-6341dbfd666c","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-08T04:10:46.094404Z","strongest_claim":"Extensive experiments on three public benchmarks (ETH-UCY, SDD, and NBA) demonstrate that our method consistently outperforms strong single-model and ensemble baselines, achieving an average improvement of 10.5%, showcasing the effectiveness of scene-aware selective learning.","one_line_summary":"SceneSelect discovers a latent scene taxonomy through clustering, trains a decoupled classifier to assign inputs, and uses a scheduling policy to dispatch to optimal expert trajectory predictors, reporting 10.5% average gains over baselines on ETH-UCY, SDD, and NBA.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That unsupervised clustering on geometric and kinematic features produces a stable, generalizable latent scene taxonomy, and that the decoupled classification and scheduling modules enable seamless integration with off-the-shelf models and new datasets without joint retraining or performance loss.","pith_extraction_headline":"SceneSelect classifies input scenes from geometry and motion features then routes each trajectory to the best expert predictor."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.24514/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-21T06:39:13.159730Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T22:02:31.665130Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"9b0ef6149bff81ba1b9fc6b3b9855e6efa1a19400975a09c0bf843445b155a06"},"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"}