[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

When to break anchor?

Posted By: Timothy Chow
Date: Monday, 29 June 2015, at 11:57 p.m.

In Response To: When to break anchor? (Dmitriy Obukhov)

Currently, my general approach to this type of problem (and most other types of problems, for that matter) is:

1. Collect lots of examples from my training matches against the bot, sorting them into categories.

2. Study the examples as a group and try to get a general feeling for the factors involved.

The drawback of this approach is that it takes a very long time to build up a sufficiently large library of positions to be of real use. I've been doing this for about two years now, and although I've been reaping some benefits along the way, it's only about now that I feel that I have enough positions to really start benefiting from all the groundwork I've been laying. For example, a particular class of problems that I have trouble with is breaking the 18pt anchor when I roll 5x (and own my midpoint). I now have a few dozen positions of this type, which IMO is around the right number to start really training my carbon-based neural net.

The process can of course be supplemented by generating one's own variants. The limitation of this approach is that it tends not to produce enough variety. Nature is better at generating variants than I am.

Messages In This Thread

 

Post Response
Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:
 

 

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums is maintained by Stick with WebBBS 5.12.