Re: Novelty search in competitive coevolution thesis incoming
- Hi Drew, sounds very interesting, good luck on finishing your thesis and I look forward to seeing the results.
--- In email@example.com, Drew Kirkpatrick <drew.kirkpatrick@...> wrote:
> Just wanted to give a heads up that I've finished my experiments and
> analysis of using NCD for novelty search in competitive coevolution,
> and even with my very basic stats analysis it appears to work.
> I'm blowing off the next few weeks of work, flown in family to help
> watch the baby, and I'm scrambling to write my thesis now as I'm under
> a big time crunch for my defense April 20th. I'm hoping in the next
> few weeks I can get the paper up here, and get all of the source code
> posted (it's all C++ written for Linux boxes).
> I have 10 conditions in my experiment, which is basically a repeat of
> Ken's competitive coevolution complexification experiment from a few
> years back. The conditions are standard objective neat, random search,
> and then 4 different methods of novelty search using NCD, repeated
> with and without minimal criteria. A few of these novelty search
> approaches stand out in particular in their performance in finding
> dominant strategies and objective performance.
> Much thanks to Ken and the rest of you for answering my questions the
> past year or so, this users group has provided much useful guidance.
> Hopefully the paper doesn't turn out as badly as I think it will, and
> is of interest to some here. Too much time coding, not enough time