{"paper":{"title":"Generalized Functional ANOVA in Closed-Form: A Unified View of Additive Explanations","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Hilbert space methods yield an explicit Riesz basis for generalized functional ANOVA on continuous dependent inputs","cross_cats":["cs.LG","math.ST","stat.TH"],"primary_cat":"stat.ML","authors_text":"Baptiste Ferrere, Fabrice Gamboa, Jean-Michel Loubes, Nicolas Bousquet","submitted_at":"2026-05-18T13:56:10Z","abstract_excerpt":"The functional ANOVA, or Hoeffding decomposition, provides a principled framework for interpretability by decomposing a model prediction into main effects and higher-order interactions. For independent inputs, this classical decomposition is explicit. It is closely connected to SHAP values, generalized additive models, and orthogonal polynomial expansions, and therefore constitutes a fundamental tool for additive explainability. In the more general and realistic dependent setting, however, obtaining a tractable representation and estimating the decomposition from data remain challenging. In th"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"By combining Hilbert space methods with the generalized functional ANOVA, we build an explicit decomposition Riesz Basis allowing to easily compute the decomposition. Our formulation recovers the classical independent case and its associated orthogonal decomposition.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The construction assumes that the input variables are continuous and that a suitable Riesz basis exists in the underlying Hilbert space so that the decomposition remains both explicit and estimable from finite samples.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Introduces a Riesz basis for explicit generalized functional ANOVA decomposition under input dependence and an associated data-driven estimation procedure.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Hilbert space methods yield an explicit Riesz basis for generalized functional ANOVA on continuous dependent inputs","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"3e3521a6aed4090588fb876f967c71c0b2f5b74dc8be3750fcc01e680eda0465"},"source":{"id":"2605.18422","kind":"arxiv","version":1},"verdict":{"id":"986f9a1b-f4ad-468f-bc55-22652f3af5e7","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T23:53:39.247352Z","strongest_claim":"By combining Hilbert space methods with the generalized functional ANOVA, we build an explicit decomposition Riesz Basis allowing to easily compute the decomposition. Our formulation recovers the classical independent case and its associated orthogonal decomposition.","one_line_summary":"Introduces a Riesz basis for explicit generalized functional ANOVA decomposition under input dependence and an associated data-driven estimation procedure.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The construction assumes that the input variables are continuous and that a suitable Riesz basis exists in the underlying Hilbert space so that the decomposition remains both explicit and estimable from finite samples.","pith_extraction_headline":"Hilbert space methods yield an explicit Riesz basis for generalized functional ANOVA on continuous dependent inputs"},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.18422/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_title_agreement","ran_at":"2026-05-20T00:01:20.265472Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-20T00:01:14.953080Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"citation_quote_validity","ran_at":"2026-05-19T23:49:55.564072Z","status":"skipped","version":"0.1.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T23:33:27.587695Z","status":"skipped","version":"1.0.0","findings_count":0},{"name":"external_links","ran_at":"2026-05-19T23:31:34.596987Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"cited_work_retraction","ran_at":"2026-05-19T23:21:59.943990Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T23:21:58.678745Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"0abd002b1617fc922060f90eae8e20cb9b24ee4fdb8e2c32ff8e65febbaf25ef"},"references":{"count":62,"sample":[{"doi":"","year":2021,"title":"Agarwal, R., Melnick, L., Frosst, N., Zhang, X., Lengerich, B., Caruana, R., and Hinton, G. E. (2021). Neural additive models: Interpretable machine learning with neural nets.Advances in neural inform","work_id":"4d13a594-dff1-4f8d-8f67-865c1d62350e","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"I., Salaün, T., and Brunel, N","work_id":"8fa5d1ce-ecc1-4662-9781-e3df1d1392f0","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"Apley, D. W. and Zhu, J. (2020). Visualizing the effects of predictor variables in black box super- vised learning models.Journal of the Royal Statistical Society Series B: Statistical Methodology, 82","work_id":"5e6ebb58-f51c-4f87-b245-83e0e3f30589","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2018,"title":"Arzamasov, V ., Böhm, K., and Jochem, P. (2018). Towards concise models of grid stability. In 2018 IEEE international conference on communications, control, and computing technologies for smart grids ","work_id":"2e5b8026-2fcc-475d-ba7c-c77be9d2287e","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2015,"title":"Bach, S., Binder, A., Montavon, G., Klauschen, F., Müller, K.-R., and Samek, W. (2015). On pixel-wise explanations for non-linear classifier decisions by layer-wise relevance propagation. PloS one, 10","work_id":"76edb807-f995-42e6-8842-c002aa3275da","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":62,"snapshot_sha256":"bf9d542de576e7acb237aced991cf1f86fb704ce13bb21a570ac993d00e56af2","internal_anchors":2},"formal_canon":{"evidence_count":2,"snapshot_sha256":"ea06595dd2c482cb797a8c31ba155d541f22938b7bcaca55db6766cfe790e8da"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}