Home * People * Gil Fuchs
GilEFuchs.jpg

Gil E. Fuchs,
an Israeli computer scientist with a Ph.D. from University of California, Santa Cruz in 2004. At UCSC he researched along with Robert Levinson on conceptual graphs (CGs), and pattern-weight pairs (pws) to represent knowledge for search and planning. Pws were the basic representational entity of Morph III, a self learning pattern-oriented chess program [1], and were also applied to other board games such as Diplomacy [2]. Patterns are represented as conceptual position graphs, the edges being attack relationships and the nodes being pieces and important squares. Pattern weights are adjusted by temporal-difference learning in self-play, in which each pattern has its own learning rate set by simulated annealing (i.e., the more frequently a pattern is updated, the slower becomes its learning rate). Morph evaluates a position by combining the weights of all matched patterns into a single evaluation function score, and performs only a one-ply search [3].
Gil Fuchs [4]

Selected Publications

[5]

External Links


References

  1. ^ Robert Levinson, Gil Fuchs (1994). A Pattern-Weight Formulation of Search Knowledge. UCSC-CRL-94-10, CiteSeerX
  2. ^ Ari Shapiro, Gil Fuchs, Robert Levinson (2002). Learning a Game Strategy Using Pattern-Weights and Self-play. CG 2002
  3. ^ Johannes F├╝rnkranz (2000). Machine Learning in Games: A Survey. Austrian Research Institute for Artificial Intelligence, OEFAI-TR-2000-3, pdf, 5.3 Pattern-based learning systems
  4. ^ Gil E. Fuchs | Facebook
  5. ^ dblp: Gil Fuchs

What links here?


Up one level