Unbalance-Aware Deep Learning of Physical System

Abstract

The simulation of complex physical systems described by partial differential equation (PDE) is a central topic in various fields. Many training strategies for deep learning have developed for images or natural languages, but they are not necessarily suited for physical systems. A physical system demonstrates similar phenomena in most points but exhibits a drastic behavior occasionally, implying that a physical system dataset suffers from the class imbalance, whereas previous studies have rarely focused on this aspect. In this paper, we propose an imbalance-aware loss for learning physical systems, which resolves the class imbalance in a physical system dataset by focusing on the hard-to-learn parts. We evaluated the proposed loss on the PDE systems, and demonstrated that a model trained using the proposed loss outperformed the baselines by a large margin.