Vitali Petsiuk is a 2nd-year Computer Science Ph.D. student advised by Professor Kate Saenko at Boston University. He does research in Computer Vision and Deep Learning, currently working on Explainable AI, aiming to develop tools for interpretable model analysis. Vitali got his BS and MS in Computer Science from Belarusian State University, where he worked, among other things, on Graph Theory and Image Segmentation.
RISE: Randomized Input Sampling for Explanation of Black-box Models
Deep neural networks are being used increasingly to automate data analysis and decision making, yet their decision-making process is largely unclear and is difficult to explain to the end users. In this paper, we address the problem of Explainable AI for deep neural networks that take images as input and output a class probability. We propose an approach called RISE that generates an importance map indicating how salient each pixel is for the model's prediction. In contrast to white-box approaches that estimate pixel importance using gradients or other internal network state, RISE works on black-box models. It estimates importance empirically by probing the model with randomly masked versions of the input image and obtaining the corresponding outputs. We compare our approach to state-of-the-art importance extraction methods using both an automatic deletion/insertion metric and a pointing metric based on human-annotated object segments. Extensive experiments on several benchmark datasets show that our approach matches or exceeds the performance of other methods, including white-box approaches.