# Search Results

Results tagged with
Search options answers only user 1641
5 results

For questions related to policies (as defined in reinforcement learning or other AI sub-fields).

So, here's is the question: Is it true that a non-stationary policy must satisfy this condition? $$\forall i, j \in \mathbb{N}, s \in S, \pi (i, s) = \pi(j, s)$$ With your custom notation …
answered Feb 23 '19 by Dennis Soemers
Both of those policies are optimal, both reach the goal state in two steps and receive a return of $\gamma \times 1 = 0.9$, but they are clearly different policies, they choose different actions in for … Note that my language was slightly informal above when talking about "policies for the starting state". …
answered Aug 20 '18 by Dennis Soemers
And if I then have multiple different MDPs, all without any policies or anything like that, I could compare them based on those properties. … On the other hand, it can also be interesting sometimes to study multiple different policies all for the same MDP. …
answered May 24 '20 by Dennis Soemers
Here is the gradient that they are discussing in the video: \nabla_{\theta} J(\theta) \approx \frac{1}{N} \sum_{i=1}^N \left( \sum_{t=1}^T \nabla_{\theta} \log \pi_{\theta} (\mathbf{a}_{i, t} \vert …
answered Sep 6 '18 by Dennis Soemers
That would only be equivalent to a mapping from states to actions for deterministic policies, not for stochastic policies. … Assuming that our agent has access to (estimates of) value functions $Q(s, a)$ for state-action pairs, the greedy and $\epsilon$-greedy policies can be described in precisely the same way. …
answered Feb 10 '19 by Dennis Soemers