26.11.08 The Age of Analogue Networks

Currently I am trying to organize all the paperwork I have accumulated over the past year to get a perspective of where I am in my research project. As well as merely organizing and filing my work I am also trying to summarize some of the more raw materials that I have gathered. For example I am working on a list of methodologies used in AI development to optimize and rate the performance of agents, with a specific focus on AI for games.

To find a good source of research project descriptions other than the thousands of disparate papers online I went out to get “AI Game Programming Wisdom” (Edited by Steve Rabin) from the library- as it features a good collection of post-mortems of successful game AI examples (such as Halo 3).

In the library I also had a flick through the fall issue of “AI Magazine” (vol 29 no 3) and an article titled “The Age of Analogue Networks” caught my eye- This seems to be exactly the kind of thinking I have been doing in the past years, that of avoiding “digitization” of analogue network nodes (as most classic neural networks do at some point) just to allow engineers to employ classical computer science architectures. I just got a copy of this article online from here:

http://infoscience.epfl.ch/record/121372