X-Message-Number: 26101 From: Date: Thu, 28 Apr 2005 07:10:08 EDT Subject: Uploading technology (1.i.1) Uploading technology (1.i.1). A uploading project must include four elements: - A molecular scale scanner. - An image processing software able to extract informations from the scanner data. - An electronics brain simulator. - An interface system to link the brain and the world. I am mostly concernet here with the third element, but the other must not be neglected. For example the interface system must include the sensory inputs and a robot body. The picture processor must be able to see the dendrite elements, sort them out from glial cells, look at significant element such cytoskeleton molecules in dendrite spines, deduce the presence of a spine from decay products of channel receptors and so on. These products may have migrated because of both, the brownian motion and a nearby ice crystal formation for example. The main work would be done by a computer, but some image cleaning may be worked out by a neural network. Producing such a neural net and integrating into a computer could be an interesting intermediate step before looking at a full scale brain size electronics processor. I must say that I am not a specialist in computer picture processing, but I'll try to get some basic understanding of the technology. The two elements to look at are, it seems, the 3D picture using voxel (volume element), not pixel (picture element) and a block element of the neural net using a single FPGA. This elementary system would include all or nearly so what would be found in a brain quality neural net. There would be the time sharing capability, the memory on dendrite tree, the slow process on computer and so on. Producing these systems will cost a lot of money, so there must be too some intermediate market products. For exemple to recognize a partly hidden or destroyed object asks for some understanding of what is seen. In the same way, to translate speech from one language to another implies some background knowledge of the world, the society and so on. The basic problems are not very different. May be a real time translator would be a better market choice than the 3D picture interpreter... In the same way, working with voxels may be hard at first. It could be simpler ti use 2D pictures taken in different plane. The scanner itself can produce its output in this form. Yvan Bozzonetti. Content-Type: text/html; charset="US-ASCII" [ AUTOMATICALLY SKIPPING HTML ENCODING! ] Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=26101