Analogy-Making as Perception: A Computer Model (Neural Network Modeling and Connectionism)

I can't believe I waited so long to read this book -- it truly is a classic. This is the way AI should be done: by focusing on the right level of abstraction, situated above the level of neuroscience but below the level of simple input-output function mapping. Finally, a computer model that makes those first steps towards doing the same thing that people do.True believers in those original goals of artificial intelligence take heart -- this book gives new hope to a field that has come to be dominated by engineering approaches that only work in special cases like the logic behind the cruise control switch in a car. Mitchell's model provides the fluidity and flexibility that is lacking from classical machine learning techniques.

No problems, good experience.

Since AI researchers are generally engineers, they historically did what engineers do: they broke up the mind in very clear-cut divisions, one for the perception of the things out there in the world, and another, symbolically, to do "abstract cogitation".For deep reasons, this was an invalid move, but only a few could see it. Melanie surely could, for her highly original copycat project exhibits some of the best insights in Artificial Intelligence ever.AI is still so much pervaded with the wrong ideas that this book will need to take some time to make its definitive mark on the history of the field.If genuine understanding is ever to be built into a machine, understanding of the kind that Searle's gang will be forever denying, then it will come from an architecture similar to that proposed in this book.Then again, I could turn out to be wrong. But let us let History decide on this issue.