{"paper":{"title":"Architecture-Induced Recoverability Bias in Differentiable Symbolic Regression","license":"http://creativecommons.org/licenses/by/4.0/","headline":"In symbolic regression the tree architecture determines which targets gradient descent recovers, not the structure's expressiveness.","cross_cats":["cs.AI","cs.LG","cs.SC"],"primary_cat":"cs.NE","authors_text":"Chakshu Gupta, Theodore J. LaGrow","submitted_at":"2026-04-25T11:32:09Z","abstract_excerpt":"Symbolic regression aims to recover closed-form expressions from numerical data, but in differentiable symbolic regression the recovered expression depends not only on the grammar but also on the fixed architecture through which variables are routed during training. This is relevant to signal-processing settings in which closed-form models and interpretable nonlinear structure are useful. This architecture-specific effect has rarely been isolated directly, because existing comparisons often vary architecture together with operator family, grammar, or search procedure. Three depth-3 architectur"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Expressiveness guarantees that a solution exists in the search space, but not that gradient descent finds it: the most expressive structure fails on targets that a restricted alternative solves reliably.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That differences in recovery rates across the three structures are attributable to the tree architecture itself rather than to unstated details of initialization, hyperparameter choices, or the specific set of target functions and operators selected for testing.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Different fixed tree architectures in gradient-based symbolic regression produce dramatically different recovery rates, with more expressive structures sometimes failing where restricted ones succeed reliably.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"In symbolic regression the tree architecture determines which targets gradient descent recovers, not the structure's expressiveness.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"91b5389076e1a0292a8e0e4082c63ae1c1bd299215b93e3b10b6176c7304d621"},"source":{"id":"2604.23256","kind":"arxiv","version":2},"verdict":{"id":"c359dd68-3d54-4320-a8e1-c0bee968107a","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-08T06:44:22.196733Z","strongest_claim":"Expressiveness guarantees that a solution exists in the search space, but not that gradient descent finds it: the most expressive structure fails on targets that a restricted alternative solves reliably.","one_line_summary":"Different fixed tree architectures in gradient-based symbolic regression produce dramatically different recovery rates, with more expressive structures sometimes failing where restricted ones succeed reliably.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That differences in recovery rates across the three structures are attributable to the tree architecture itself rather than to unstated details of initialization, hyperparameter choices, or the specific set of target functions and operators selected for testing.","pith_extraction_headline":"In symbolic regression the tree architecture determines which targets gradient descent recovers, not the structure's expressiveness."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.23256/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-21T09:36:34.130189Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T23:17:27.128072Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"5ade97b8c75380df7ba9ff2d1205b67b7fe0a4d9db9a2333cf4aec1804ddc69a"},"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"}