{"id":"W4405907170","doi":"10.1109/tbdata.2024.3524105","title":"Data Reconstruction and Protection in Federated Learning for Fine-Tuning Large Language Models","year":2024,"lang":"en","type":"article","venue":"IEEE Transactions on Big Data","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008904543,0.0009315028,0.001294378,0.0007121806,0.001341937,0.002841529,0.002386664,0.002259762,0.001906872],"category_scores_gemma":[0.03603735,0.000706057,0.0012637,0.001066863,0.003581972,0.007387907,0.00778862,0.004554875,0.0009201628],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001530851,"about_ca_system_score_gemma":0.002483111,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00107254,"about_ca_topic_score_gemma":0.0007778943,"domain_scores_codex":[0.9894717,0.004675146,0.0007157394,0.001738215,0.002493035,0.0009060192],"domain_scores_gemma":[0.9793602,0.007160964,0.0009969601,0.0110272,0.001131213,0.0003235409],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001990489,0.0004703029,0.007694039,0.0003154887,0.0002420674,0.0009107337,0.001501295,0.4239017,0.02116396,0.2540282,0.006686278,0.2810954],"study_design_scores_gemma":[0.00006836686,0.0001294105,0.0003953212,0.00004191916,0.00002373842,0.0003248809,0.0001412577,0.8011526,0.02117624,0.1740611,0.00244796,0.00003714413],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.04237363,0.0002391248,0.9522457,0.0009203559,0.00005698984,0.00009518514,0.0001356199,0.002358465,0.001575014],"genre_scores_gemma":[0.9049994,0.0001146163,0.09248409,0.0004988169,0.00003478405,0.0001594987,0.0001737122,0.0001571376,0.001378053],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.008904543,"threshold_uncertainty_score":0.04709232,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1394083824470668,"score_gpt":0.3164687905736269,"score_spread":0.1770604081265601,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}