Dynamic proximal policy optimization: Enhancing PPO with adaptive entropy and smooth clipping | Synapse