Debiased-CAM to mitigate systematic error with faithful visual explanations of machine learning

Abstract

Model explanations such as saliency maps can improve user trust in AI by highlighting important features for a prediction. However, these become distorted and misleading when explaining predictions of images that are subject to systematic error (bias). Furthermore, the distortions persist despite model fine-tuning on images biased by different factors (blur, color temperature, day/night). We present Debiased-CAM to recover explanation faithfulness across various bias types and levels by training a multi-input, multi-task model with auxiliary tasks for explanation and bias level predictions. In simulation studies, the approach not only enhanced prediction accuracy, but also generated highly faithful explanations about these predictions as if the images were unbiased. In user studies, debiased explanations improved user task performance, perceived truthfulness and perceived helpfulness. Debiased training can provide a versatile platform for robust performance and explanation faithfulness for a wide range of applications with data biases.

Ensemble Learning, Deep Learning-Based and Molecular Descriptor-Based Quantitative Structure–Activity Relationships

“…The relationship between bias and variance follows the relationship that if one side wins, the other side will not win, i.e., the bias-variance tradeoff [25]. Bias in ML and statistics model prediction refers to the difference between the predicted and true values, that is the bias error, arising from incorrect model assumptions, where bias is the error due to the simplification of the actual problem [26]. For example, linear regression simplifies the problem.…”