English

Data-Adaptive Discriminative Feature Localization with Statistically Guaranteed Interpretation

Machine Learning 2022-11-21 v1 Artificial Intelligence Machine Learning Applications Methodology

Abstract

In explainable artificial intelligence, discriminative feature localization is critical to reveal a blackbox model's decision-making process from raw data to prediction. In this article, we use two real datasets, the MNIST handwritten digits and MIT-BIH Electrocardiogram (ECG) signals, to motivate key characteristics of discriminative features, namely adaptiveness, predictive importance and effectiveness. Then, we develop a localization framework based on adversarial attacks to effectively localize discriminative features. In contrast to existing heuristic methods, we also provide a statistically guaranteed interpretability of the localized features by measuring a generalized partial R2R^2. We apply the proposed method to the MNIST dataset and the MIT-BIH dataset with a convolutional auto-encoder. In the first, the compact image regions localized by the proposed method are visually appealing. Similarly, in the second, the identified ECG features are biologically plausible and consistent with cardiac electrophysiological principles while locating subtle anomalies in a QRS complex that may not be discernible by the naked eye. Overall, the proposed method compares favorably with state-of-the-art competitors. Accompanying this paper is a Python library dnn-locate (https://dnn-locate.readthedocs.io/en/latest/) that implements the proposed approach.

Keywords

Cite

@article{arxiv.2211.10061,
  title  = {Data-Adaptive Discriminative Feature Localization with Statistically Guaranteed Interpretation},
  author = {Ben Dai and Xiaotong Shen and Lin Yee Chen and Chunlin Li and Wei Pan},
  journal= {arXiv preprint arXiv:2211.10061},
  year   = {2022}
}

Comments

27 pages, 11 figures

R2 v1 2026-06-28T06:11:14.283Z