|
Memorial University, Canada
by:
Dr. Simon Harding
Dr. Wolfgang Banzhaf
|
Abstract:
Genetic programming (GP) is a powerful form of machine
learning. However, it relies on the evaluation of vast
numbers - potentially millions - of candidiate solutions.
This evaluation is a bottleneck. Luckily, in general we
find that these problems are relatively easy to
parallelize. Using GPUs it is possible to significantly
reduce this computational bottleneck and start to address
harder and more interesting machine learning problems.
We have developed a sophisticated GP system that uses a
cluster of GPU equipped machines. This system allows for
the efficient processing of very large datasets containing
over 10 million rows and several hundred megabytes in size.
|