{"id":"W4389584550","doi":"10.1109/tai.2023.3340982","title":"Manipulation Attacks on Learned Image Compression","year":2023,"lang":"en","type":"article","venue":"IEEE Transactions on Artificial Intelligence","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Calgary; University of Waterloo","funders":"Army Research Office; Science and Technology Program of Hubei Province; National Natural Science Foundation of China","keywords":"Computer science; Image compression; Lossy compression; Lossless compression; Artificial intelligence; JPEG; Deep learning; Robustness (evolution); Image quality; Computer vision; Data compression; Computer engineering; Image processing; Image (mathematics)","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":"codex-gemma-dda1882f352a","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0004102957,0.000228825,0.0001860513,0.0004264979,0.0006636215,0.0002106914,0.0007368617,0.000122565,0.0001771311],"category_scores_gemma":[0.00004416888,0.0002293031,0.0001284209,0.001279318,0.00008892947,0.0005162813,0.00001078954,0.0005972083,0.004283186],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000891245,"about_ca_system_score_gemma":0.00003610845,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004817877,"about_ca_topic_score_gemma":0.00002434892,"domain_scores_codex":[0.9978626,0.0001868859,0.0004034375,0.0006276617,0.0005217704,0.0003976741],"domain_scores_gemma":[0.9986117,0.0004115475,0.0001115656,0.0006630956,0.00008942498,0.000112665],"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":[0.00004862465,0.00009267671,0.000002030658,0.000006268539,0.000008463547,0.00002017014,0.0004736466,0.70184,0.005144675,0.009923,0.00009012566,0.2823503],"study_design_scores_gemma":[0.00003972112,0.0001496552,0.00007992277,0.00005185313,0.000006846463,0.000003563044,0.0001172553,0.861055,0.1284387,0.009548121,0.0002700329,0.000239338],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.008920759,0.000002278068,0.9852665,0.001563337,0.001926516,0.0002301667,0.000003478486,0.0009975887,0.001089386],"genre_scores_gemma":[0.9896033,0.00001745663,0.009473401,0.0001917413,0.0001069413,0.00002973062,0.000003862656,0.0000279572,0.0005455588],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9806826,"threshold_uncertainty_score":0.9964921,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1014077847971648,"score_gpt":0.357783726830343,"score_spread":0.2563759420331782,"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."}}