[Computer-go] Fwd: News on Tromp-Cook ?

[Aja]

Hi Fuming,

I agree with your understanding. As for the naming of UCT, I think it's fine, because we can just view such formula as assigned with zero coefficient of the exploration term.

Aja

  ----- Original Message ----- 
  From: Fuming Wang 
  To: Aja 
  Cc: computer-go at dvandva.org 
  Sent: Sunday, January 02, 2011 10:35 PM
  Subject: Re: [Computer-go] Fwd: News on Tromp-Cook ?


  Hi Aja,



  On Sun, Jan 2, 2011 at 1:29 AM, Aja <ajahuang at gmail.com> wrote:

    Hi Fuming,
    Â 
    C*RAVE+(1-C)*UCT
    Â 
    C is computed dynamically in search, but not set to a fixed value. Maybe you mean UCT_C,
    Â 
    UCT=UCT_mean+UCT_C*exploration_term
    Â 
    What Petr and Olivier do, I think, is set UCT_C to 0, to disable the exploration_term, not the weight of RAVE. 



  Without the exploring term, the UCT is just mean win rate, so there's no point in calling it UCT or UCB. Basically, what people have been saying is that currently the tree search is based on combination of sequence dependent rate (average win rate) and sequence independent/almost independent (rave rate) instead of combination of exploitation (win rate) and exploration (UCB term). Is this understanding close?

  Fuming 


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://computer-go.org/pipermail/computer-go/attachments/20110102/553defd9/attachment.html>