{"paper":{"title":"SMART Fine-tuning Factor Augmented Neural Lasso","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Fine-tuning the factor-augmented neural Lasso yields minimax-optimal excess risk bounds and statistical acceleration over single-task learning when relative sample sizes and function complexities align in high-dimensional nonparametric reg","cross_cats":["cs.LG","stat.ME"],"primary_cat":"stat.ML","authors_text":"Cheng Gao, Jianqing Fan, Jinhang Chai, Qishuo Yin","submitted_at":"2026-04-14T05:01:18Z","abstract_excerpt":"Fine-tuning is a widely used strategy for adapting pre-trained models to new tasks, yet its methodology and theoretical properties in high-dimensional nonparametric settings with variable selection have not yet been developed. We propose a source-model-augmented residual tuning (SMART) framework, which incorporates the pre-trained source model as an augmented feature into the target learner and estimates only the residual target-specific component. The approach is widely applicable, from parametric and sparse models to neural networks and blackbox machine learning models. We focus on the devel"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We derive minimax-optimal excess risk bounds for the fine-tuning FAN-Lasso, characterizing the precise conditions, in terms of relative sample sizes and function complexities, under which fine-tuning yields statistical acceleration over single-task learning.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The target function admits a residual fine-tuning decomposition as a transformation of a frozen source function plus other variables, combined with a low-rank factor structure adequately capturing high-dimensional dependent covariates.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"FAN-Lasso uses low-rank factor structures and a residual fine-tuning decomposition to enable transfer learning and variable selection in high-dimensional nonparametric regression, delivering minimax-optimal excess risk bounds under conditions on sample sizes and function complexity.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Fine-tuning the factor-augmented neural Lasso yields minimax-optimal excess risk bounds and statistical acceleration over single-task learning when relative sample sizes and function complexities align in high-dimensional nonparametric reg","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"98cd2638b89f3e0c48f9e13615f5d2fa0963726b000125350bfa787305edf58e"},"source":{"id":"2604.12288","kind":"arxiv","version":2},"verdict":{"id":"94061665-dfa6-42d1-b24f-e7122688538b","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T15:41:45.219528Z","strongest_claim":"We derive minimax-optimal excess risk bounds for the fine-tuning FAN-Lasso, characterizing the precise conditions, in terms of relative sample sizes and function complexities, under which fine-tuning yields statistical acceleration over single-task learning.","one_line_summary":"FAN-Lasso uses low-rank factor structures and a residual fine-tuning decomposition to enable transfer learning and variable selection in high-dimensional nonparametric regression, delivering minimax-optimal excess risk bounds under conditions on sample sizes and function complexity.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The target function admits a residual fine-tuning decomposition as a transformation of a frozen source function plus other variables, combined with a low-rank factor structure adequately capturing high-dimensional dependent covariates.","pith_extraction_headline":"Fine-tuning the factor-augmented neural Lasso yields minimax-optimal excess risk bounds and statistical acceleration over single-task learning when relative sample sizes and function complexities align in high-dimensional nonparametric reg"},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.12288/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"}