PE-MHL: Physics-Encoded Modular Hybrid Layers

Arxiv CS.LG (machine learning)

The paper introduces PE-MHL, a hybrid model architecture that integrates physics-based equations into neural layers for scalable learning of complex systems. It shows improved accuracy and interpretability in control applications compared to purely data-driven approaches.

models

Arxiv CS.LG (machine learning)·Ismail HassaballaMircea Lazar

7 days ago