試験
通常
試験時間: 00:00問題時間: 00:00
問題一覧
問題 44 / 65
Which method is suitable when you want to quantitatively understand, at the whole-model level, the contribution of each feature of a deep learning model to its predictions?
①LIME (local interpretation)
②SHAP values (contributions based on Shapley values)
③Dimensionality reduction by principal component analysis (PCA)
④Grad-CAM (visual explanation)
⑤Simple feature correlation coefficient