{"id":"W4392920822","doi":"10.21203/rs.3.rs-3648954/v1","title":"LRCM: Enhancing Adversarial Purification through Latent Representation Compression","year":2024,"lang":"en","type":"preprint","venue":"Research Square","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"","funders":"Guangdong Power Grid Company; Canadian Institute for Advanced Research","keywords":"Adversarial system; Representation (politics); Compression (physics); Artificial intelligence; Computer science; Pattern recognition (psychology); Political science; Materials science; Law; Composite material","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002113677,0.001219113,0.001166037,0.0007959722,0.0004554685,0.001138656,0.001865448,0.00182883,0.005650939],"category_scores_gemma":[0.00818395,0.0005410678,0.0007144453,0.0009026698,0.001456733,0.00234727,0.003769632,0.003011203,0.002484761],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004930376,"about_ca_system_score_gemma":0.0008963391,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001624594,"about_ca_topic_score_gemma":0.001954084,"domain_scores_codex":[0.9980041,0.0007744461,0.00005826271,0.0003226448,0.0007032357,0.000137296],"domain_scores_gemma":[0.997049,0.001362079,0.0001665056,0.00102707,0.000286993,0.0001083434],"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":[0.000411223,0.0002505414,0.0005962556,0.0002189739,0.0001314975,0.0002167638,0.0001178561,0.4737655,0.03035817,0.0683599,0.01791208,0.4076612],"study_design_scores_gemma":[0.00001319812,0.00003394063,0.00007402958,0.000008931115,0.000008348056,0.00005763357,0.000005736222,0.9769633,0.006824108,0.01427448,0.001724869,0.00001136583],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.004240203,0.0002318328,0.9921936,0.0002399306,0.00007257588,0.00003296512,0.00007420569,0.001533241,0.00138156],"genre_scores_gemma":[0.3683205,0.0005664509,0.6121663,0.0008728564,0.0003278052,0.0001755228,0.0008478584,0.001129598,0.01559313],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005650939,"threshold_uncertainty_score":0.01890427,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0958234845702851,"score_gpt":0.4326338984710349,"score_spread":0.3368104139007498,"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."}}