1 paper · 1 filter
Chenyu Xue, Fan Wang, Yuanzhuo Zhu +4
Deploying reliable deep learning techniques in interdisciplinary applications needs learned models to output accurate and (even more importantly) explainable predictions. Existing…