The study uses the idea of the extreme parallel to solve the BP neural network classification. Modification of the weights is not the traditional method which is to modify the connection weights between neurons repeatedly, but to find a group of weights in all possible weights combinations. The groups of weights are suitable for the relationship of the ideal input and the ideal output. Therefore, the model has some advantages compared with the traditional serial model in time miscellaneous. In the actual DNA computing, we also associate the coding problem with the model design. The coding problem is an important issue worthy to study in the DNA computing. There are many factors affecting the coding. The coding in this study is made when certain factors are overlooked.
[1]
Clifford R. Johnson,et al.
Solution of a 20-Variable 3-SAT Problem on a DNA Computer
,
2002,
Science.
[2]
Lloyd M. Smith,et al.
DNA computing on surfaces
,
2000,
Nature.
[3]
E. Shapiro,et al.
An autonomous molecular computer for logical control of gene expression
,
2004,
Nature.
[4]
Darko Stefanovic,et al.
Deoxyribozyme-based logic gates.
,
2002,
Journal of the American Chemical Society.
[5]
Qi Cheng,et al.
Linear Self-Assemblies: Equilibria, Entropy and Convergence Rates
,
2003
.
[6]
Junghuei Chen,et al.
A DNA-Based Memory with In Vitro Learning and Associative Recall
,
2003,
DNA.