gymR

• CartPole_v0 example: [https://gym.openai.com/docs/]
• What if the stochastic approximation method is applied to reinforcement learning problems such as CartPole?
• Here is result.

1. Finite-difference stochastic approximation algorithm (FDSA)

---

2. Simultaneous pertubation stochastic approximation algorithm (SPSA)

---

3. Adaptive SPSA (2SPSA)