[Computer-go] parameter optimization (clop & others)

Olivier Teytaud teytaud at lri.fr
Thu Feb 7 12:01:32 PST 2013


Thanks to you both for feedback;

The CLOP paper has references to papers with these bounds (Chen-1988 for
> the proof that it is not possible to do better. IIRC, some variations of
> stochastic gradient have been proved to match the bound, so the bound is
> tight).
>
> Yes, Chen's bound is the main reference for bounds in the differentiable
case;
and the Kiefer-Wolfowitz algorithm is the main reference for showing that
the bound is matched by some algorithm.

These papers close the question for this class of functions, at least if

- we don't try to know the dependency in the number of parameters
    (the constant term in the linear slope in log-log term depends on the
number of parameters,
    this might be relevant for Brian's remark on the number of parameters).

- we don't want to restrict the assumptions

Best regards,
Olivier
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://computer-go.org/pipermail/computer-go/attachments/20130207/c734284c/attachment.html>


More information about the Computer-go mailing list