Alpha-Mini: Minichess Agent with Deep Reinforcement Learning