Friday, 1 October 2004

Woo! Good results, finally.

Mood: happy, relieved

In the last few days, I've been running some function approximation experiments that, in a nutshell, have given me exactly the results I needed!

I wasn't happy with the paper that I submitted to Complex 2004, because I thought I had a good idea but it just didn't work. The results I got from the experiments I included in that paper were statistically inconclusive and I was left grasping at straws to justify the worth of the paper.

Well, I've done two more experiments based on what I've learned recently, and both of them have given excellent results!
I couldn't ask for more conclusive data - it shows that (for fully specified fuzzy logic rulebases) multi-dimensional representations are better than single-dimensional representations. That statement comes with an iron-clad statistical significance of 0.001 - for anyone who doesn't know stats, anything under 0.05 is acceptable and lower is better.

This comes at the perfect time, because I have the opportunity to edit my paper to include the new results, and submit the new version as the final submission. I am quite satisfied with the way things have turned out. My paper is worthwhile, and my intuition has been vindicated :D

1 comment:

Anonymous said...

friday, 1 october, 2004 18:00 MAT

good job nick!! behold the power of the purple line on your graph!

,` ) melanie