{"id":"W4307168624","doi":"10.1190/geo2021-0470.1","title":"Hierarchical transfer learning for deep learning velocity model building","year":2022,"lang":"en","type":"article","venue":"Geophysics","topic":"Seismic Imaging and Inversion Techniques","field":"Earth and Planetary Sciences","cited_by":22,"is_retracted":false,"has_abstract":true,"ca_institutions":"Polytechnique Montréal; Institut National de la Recherche Scientifique","funders":"Fonds de recherche du Québec – Nature et technologies; Natural Sciences and Engineering Research Council of Canada; Society of Economic Geologists Foundation","keywords":"Computer science; Benchmark (surveying); Algorithm; Artificial neural network; Deep learning; Similarity (geometry); Mean squared error; Model building; Convergence (economics); Artificial intelligence; Transfer of learning; Salient; Inverse; Inversion (geology); Measure (data warehouse); Data mining; Mathematics; Image (mathematics); Geology; Geometry; Physics","routes":{"ca_aff":true,"ca_fund":true,"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":[],"consensus_categories":[],"category_scores_codex":[0.0003173818,0.0001064374,0.0001321018,0.00005557086,0.0009916109,0.00003594904,0.0001963314,0.00002852056,0.0002716242],"category_scores_gemma":[0.00002797401,0.0001100941,0.0001015264,0.0001604131,0.00004651855,0.0001248163,0.00002412204,0.0005729811,0.00001484669],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001032097,"about_ca_system_score_gemma":0.0000378208,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0004379468,"about_ca_topic_score_gemma":0.000002392716,"domain_scores_codex":[0.9989994,0.00008951843,0.0001243366,0.0002418992,0.0002365423,0.0003083064],"domain_scores_gemma":[0.9996686,0.0001327769,0.00002627766,0.00008500966,0.00002406488,0.00006329555],"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":[0.00004902749,0.00001445229,0.003544179,0.00001500669,0.00001088561,0.000002059261,0.0007639896,0.6757017,0.00026302,0.0008166626,0.0005795094,0.3182395],"study_design_scores_gemma":[0.0001722192,0.0001538185,0.0007814104,0.000003099386,0.00001048964,0.00000338002,0.0002189417,0.9557087,0.000374719,0.01177732,0.0306402,0.0001556765],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7152594,0.00009670223,0.2815455,0.0003971495,0.0001722663,0.0001715016,0.00002273397,0.0003128792,0.002021897],"genre_scores_gemma":[0.9899914,0.00001322831,0.008231733,0.0008499891,0.00008680749,0.000008928549,0.0001238257,0.000007707368,0.0006863313],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3180839,"threshold_uncertainty_score":0.762677,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01517218323638002,"score_gpt":0.2197240971333034,"score_spread":0.2045519138969234,"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."}}