PROSPECTS FOR LARGE-SCALE NEURAL NETWORK COMPUTATION USING DNA MOLECULES

摘要

As first pointed out by Adleman in 1994, Watson-Crick hybridization of pairs of complementary DNA strands makes possible a representation of highly parallel selective operations that could allow one to do computations using molecules. However, small departures from the ideal selectivity of DNA hybridization will inevitably lead to significant undesired pairings of strands and thus to difficulties in schemes for implementing large scale Boolean functions using DNA. In this paper we review our finding that large-scale fault-tolerant computation may be achieved using neural networks represented in DNA and suggest some applications in the field of DNA sequence determination and gene expression.