From weng@cse.msu.edu Tue Feb 15 14:24:21 2000 From: Dr John J Weng To: wdl@cse.msu.edu Subject: WDL Dis #4 Tony, Thank you very much for asking ``reference to the neuroscience literature concerning this rewiring of visual to auditory cortex in animals and their ability to relearn the mappings.'' Prof. Mriganka Sur at MIT is better suited to provide this information since what I was referring to is a series of cross-modality plasticity work that his group has been doing. Mriganka, could you please give a few references? I have some references of his with me but I am not sure that they are his most favorites. A couple of months ago, Mriganka told me their recently new findings but they have not been published yet (submitted). I wish that he can tell us some detail about them at the workshop. :-) In the book ``Rethinking Innateness'' by Jeffrey Elman, et al. there is a survey of ``Redirecting input'' (pp. 272 - 275) within the section ``Plasticity.'' Prof. Kim Plunkett who will give a review talk at our workshop is a coauthor of that book. Some AI people may feel "machines do not have to work like humans." I made a similar comment 15 years ago when I was a student attending Narendra Ahuja's computer vision class at UI. Well, that comment is true since any AI researcher can choose not to. But the AI fields are facing tremendous challenges in robotics/vision/speech/pattern recognition/language/decision making ... For example, Prof. Steve Levison wrote in his submission to WDL: ``The vast majority of participants argued that the automatic speech recognition problem will be solved very soon by incremental improvements to existing techniques. Perhapes they are correct. We think not.'' (Prof. Levison used to work at AT&T Bell Labs on speech recognition and language recognition. He is now a professor at UI. He will give a talk at our workshop.) The human brain appears to use general-purpose self-organization methods to develop all these capabilities through the process of automatic development. It appears that the brain's ways are more promising than current engineering approaches (thank our brain). On the other hand, in psychology and neuroscience, the way how brain works is very difficult to understand without identifying and understanding these self-organization principles. John