Advantage Actor Critic (A2C)
Hugging Face Blog
Advantage Actor-Critic (A2C) combines policy-based and value-based methods to reduce variance in reinforcement learning by using both an Actor that selects actions and a Critic that evaluates them. The method uses the advantage function to measure how much better an action is compared to the average value of a state, which can be estimated using temporal difference error instead of requiring separate Q and V functions. The resulting algorithm trains faster and more stably than Reinforce, enabling agents to learn complex tasks like bipedal walking in robotic simulations.