next up previous
Next: Importance Sampling Up: No Title Previous: No Title

   
Evaluating One Policy With Another

Until now we discussed the case we have policy $\pi$ and need to evlaute its value $V^{\pi}$.
Now we look at the case where we have two policies: $\pi_{1}$,$\pi_{2}$. We have samples of $\pi_{1}$ and we need to evaluate $V\pi_{2}$.


 

Yishay Mansour
2000-01-07