Enhancing Gate Fidelity and Speed Through AI-Driven Optimization for Hubbard Model Computation