TY - JOUR T1 - Dying ReLU and Initialization: Theory and Numerical Examples AU - Lu , Lu AU - Shin , Yeonjong AU - Su , Yanhui AU - Em Karniadakis , George JO - Communications in Computational Physics VL - 5 SP - 1671 EP - 1706 PY - 2020 DA - 2020/11 SN - 28 DO - http://doi.org/10.4208/cicp.OA-2020-0165 UR - https://global-sci.org/intro/article_detail/cicp/18393.html KW - Neural network, Dying ReLU, Vanishing/Exploding gradient, Randomized asymmetric initialization. AB -

The dying ReLU refers to the problem when ReLU neurons become inactive and only output 0 for any input. There are many empirical and heuristic explanations of why ReLU neurons die. However, little is known about its theoretical analysis. In this paper, we rigorously prove that a deep ReLU network will eventually die in probability as the depth goes to infinite. Several methods have been proposed to alleviate the dying ReLU. Perhaps, one of the simplest treatments is to modify the initialization procedure. One common way of initializing weights and biases uses symmetric probability distributions, which suffers from the dying ReLU. We thus propose a new initialization procedure, namely, a randomized asymmetric initialization. We show that the new initialization can effectively prevent the dying ReLU. All parameters required for the new initialization are theoretically designed. Numerical examples are provided to demonstrate the effectiveness of the new initialization procedure.