Computational Backgammon

My attempt to build a neural network-based backgammon player

Saturday, February 18, 2012

Player 3.2 1-ply benchmark

I ran my 1-ply version of Player 3.2 through the GNUbg benchmarks: Contact ER 11.0, Crashed ER 10.8, Race ER 1.67. That compares favorably to the 0-ply scores of 14.0, 12.6, and 2.08, but still worse than the 0-ply GNUbg scores.

Posted by Mark Higgins
Email ThisBlogThis!Share to XShare to FacebookShare to Pinterest

No comments:

Post a Comment

Newer Post Older Post Home
Subscribe to: Post Comments (Atom)

Pages

  • Home
  • How to build a backgammon bot

Blog Archive

  • ►  2023 (4)
    • ►  December (3)
    • ►  November (1)
  • ►  2021 (9)
    • ►  July (4)
    • ►  June (5)
  • ▼  2012 (71)
    • ►  August (2)
    • ►  June (1)
    • ►  May (1)
    • ►  April (10)
    • ►  March (16)
    • ▼  February (12)
      • Testing a new contact/crashed input: hit & cover
      • Player 3.2 1-ply benchmark
      • Bug in benchmark calculation: fixed
      • Player 3.2q: a new filter strategy
      • Player 3.2: adding a crashed network
      • Player 3.1: expanded race inputs
      • Player 3: training with supervised learning
      • GNUbg 0-ply benchmarks reference
      • Supervised learning using GNUbg training databases
      • GNUbg benchmark results
      • A better benchmark: the GNUbg benchmark database
      • Evaluating a "crashed" network
    • ►  January (29)
  • ►  2011 (30)
    • ►  December (3)
    • ►  October (3)
    • ►  September (4)
    • ►  August (15)
    • ►  July (5)

Followers

About Me

Mark Higgins
View my complete profile
Simple theme. Powered by Blogger.