Feature Attribution Method
Feature attribution methods aim to explain the predictions of complex machine learning models by assigning importance scores to input features. Current research focuses on improving the robustness and reliability of these methods, particularly for deep learning models (including transformers and convolutional neural networks), addressing issues like inconsistency across different techniques and sensitivity to perturbations. This work is crucial for building trust in AI systems across various applications, from industrial process optimization to healthcare diagnostics, by enhancing transparency and facilitating better understanding of model behavior.
Papers
October 24, 2023
October 19, 2023
October 9, 2023
September 6, 2023
August 17, 2023
July 12, 2023
July 7, 2023
July 3, 2023
June 19, 2023
April 25, 2023
February 24, 2023
January 5, 2023
December 24, 2022
December 22, 2022
December 9, 2022
November 29, 2022
November 23, 2022
May 19, 2022
May 6, 2022