Dual Cone Gradient Descent for Training Physics-Informed Neural Networks Y oungsik Hwang Artificial Intelligence Graduate School UNIST hys3835@unist.ac.kr Dong-Y oung Lim

Oct-10-2025, 13:46:04 GMT–Neural Information Processing Systems 

In this paper, we identify that PINNs can be adversely trained when gradients of each loss function exhibit a significant imbalance in their magnitudes and present a negative inner product value.