AIFT Seminar by Prof. Minghua CHEN
Prof. Minghua Chen presented a seminar on using machine learning for real-time constrained optimization on 5 Sep 2023. By using neural networks, the approach achieves faster solutions for time-critical applications like autonomous driving and high-frequency trading. He introduced two methods, predict-and-reconstruct and homeomorphic projection, to ensure NN solution strictly satisfies the equality and inequality constraints. It showed that homeomorphic projection outperforms existing methods in solution feasibility and run-time complexity, while maintaining similar optimality loss.