The presentation is going to summarize a range of my personal ideas regarding creation of a molecular computer, published on the web site http://milk.nevzorov.net, with a purpose to find people and organizations willing to help me to start up the project.
The proposed molecular computer is expected to have architecture of a 3D neural network, consisting of artificial cells, each 20-50 nm in diameter. Those cells will be able to stay in "resting" or "excited" states. Once a cell gets excited, it will affect its neighbors by causing or not causing their excitation. Every cell will get signals from its neighbors, and, if it gets excited, it will affect its other neighbors and "memorize" in the "short" memory the fact that it was recently excited. The process of excitation transfers will result in propagation and processing of signals in the whole system. When the wave of excitations reaches the outputs, the outside operator will compare the output signal with the desired one and chemically affect the cells that were recently excited in a way that will reinforce or disrupt their responses they exhibited during the last lesson. That feedback will transfer the information from the "short" memory into the "long" one. As a result of a series of lessons, the input weights of appropriate "cells" will be adjusted so that the whole system will be able to perform logical tasks of increasing complexity in a way that makes it more similar to a brain rather than conventional computer.
Structures of the model compounds and their reactions responsible for the controllable self-assembly of the molecules into organelles, organelles into the cells, and cells into the whole computer will be given. Basic reactions and processes corresponding to different stages of information processing on all levels will be presented as well.