Re: computer-go: Learning from existing games

[Markus Enzenberger <compgo@xxxxxxxxxxxxxxxxxxxxx>]

On Thursday 16 January 2003 05:06, Erik van der Werf wrote:

> Do you train on-line or in batch style (e.g., several games before a
> weight update)?
>
> I think algorithms for faster weight update only work well with
> stable gradient information. On-line learning may not provide good
> enough gradient information.

I tried batch update, but it gave worse results.
online training with TD(0) seemed to have a stabilizing effect
on the training if you go backwards through the game
and update the weights once per position.

- Markus