Data-driven forward-inverse problems for Yajima–Oikawa system using deep learning with parameter regularization

We investigate data-driven forward-inverse problems for Yajima–Oikawa (YO) system by employing two technologies to improve the performance of deep physics-informed neural network (PINN), namely neuron-wise locally adaptive activation functions and L2 norm parameter regularization. Indeed, we not only recover three different forms vector rogue waves (RWs) under distinct initial–boundary value conditions in forward problem YO system, including bright–bright RWs, intermediate–bright RWs dark–bright but also study inverse using training data with noise intensity. In order deal that capacity learning unknown parameters is ideal as utilizing interference PINN functions, thus introduce regularization, which can drive weights closer origin, into functions. Then find model strategies shows amazing effect system.