I would like to announce my masters thesis called "Intelligence without hesitation". If you got the time I would like to hear your opinions on this piece of work, either via mail or via this ng directly.)
The thesis (and an abstract) is available in PDF and Postscript format from the following page:In short, the thesis applies ideas and theory from "modern" approaches to AI to a concrete robot control problem involving maze navigation. This is compared and contrasted to "traditional" approaches.
More specifically, the theis investigates ("Modern" AI):
- Connectionism (Interactivist/"Dynamicalist")
- Interactivist representations
- Virtual modularity
- Artificial neural networks
- Genetic algorithms
As opposed to ("Traditional" AI):
- Connectionism (Representationalist)
- Search algorithms (A*, breadth-first etc.)
The experiments demonstrates, among other things, that rational behavior (Ín this case choosing the correct way through the maze) does not require an internal capacity to represent and explore alternatives (as is often assumed in traditional approaches.)
I am looking forwaed to hear from you.