3
$\begingroup$

I've been working through the Sutton + Barto RL text, implementing a number of the algos + running them in the OpenAI gym. One phenomenon that I seem to come across quite regularly is that agents who, at certain points during their training appear to be making good progress towards learning a plausible state-value/state-action functions, "catastrophically forget" the insights they glean and subsequently never recover.

To make this more concrete, here's the (smoothed) reward history of my implementation of a semi-gradient expected SARSA agent with linear function approximation and binary features running on the MountainCar environment.

Reward history

Problem Details

  • The problem has a bounded, continuous state space and a discrete, 3-action action space.
  • The learner receives a reward of $-1$ at each timestep. The episode ends when the car makes its way all the way up the hill OR it reaches 200 timesteps without success.
  • Prior to learning, I tile-coded the state space using an 8x8 grid with 8 overlapping tilings, generated using Sutton's own tiles.py script.

Learner Details

  • The learner is an implementation of the semi-gradient SARSA algorithm with linear function approximation for the Q value.
  • The agent begins with all Q weights initialized to 0.
  • The learning rate for the agent was set at $\frac{1}{10} \times (\text{# tiles})^{-1}$, where $\text{# tiles}$ in this case was $8 \times 8 \times 8 = 512$
  • During learning, the agent selected its actions using an $\epsilon$-soft policy, where $\epsilon$ is set to 0.10
  • The agent's temporal discount factor is set to $\gamma = 0.95$ (though I realize it could have been 1 given the episodic nature of the task)

My Question

One possibility I want to rule out is that my implementation of the learning agent is incorrect. To that end I am curious to know whether others experience this kind of "forgetting" behavior (even on simple problems like this), and if so, how it might be reduced.

$\endgroup$
1
$\begingroup$

The same thing was happening to me with a deep Q Network on the cart-pole problem. Having a "memory" with past (S,A,R,S) sequences and sampling it to form mini batches with the new observations helped a lot to reduce catastrophic forgetting. Reducing the step size once the agent has improved a certain amount also helped.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.