Study On Neural Network Based On DNA Strand Displacement And Its Application

As an emerging research field,biocomputing holds the promise of surpassing traditional computing and has become an important direction of study.DNA computing has gained significant attention in this field due to its highly parallel nature,high storage density,and biocompatibility.In recent years,DNA strand displacement techniques have been widely employed in DNA computing for various complex computational tasks.This paper aims to construct a winner-takes-all(WTA)neural network based on DNA strand displacement reactions and utilize it for the classification and recognition of printed Chinese numerals.The main research contributions of this paper are as follows:Firstly,this paper introduces the logical structure of the WTA neural network and provides a detailed explanation of the process of constructing the WTA network using DNA strand displacement reactions.Furthermore,this paper demonstrates the process of performing WTA neural network computations using DNA strand displacement reactions through simulation experiments and fluorescence experiments.Compared to traditional linear threshold circuits,WTA circuits demonstrate stronger computational capabilities and require fewer molecules for implementation.Additionally,they exhibit a high degree of cascadeability and modularity.Subsequently,this paper demonstrates a two-classification pattern recognition and a nine-classification pattern recognition of the printed Chinese numeral patterns in a simulation experiment.To address the issue of limited datasets for printed Chinese numerals,this paper introduces Generative Adversarial Networks(GANs)to expand the original dataset.For the more complex task of direct nine-classification,this paper employs a grouping recognition method to simplify the original problem and utilizes Monte Carlo methods to determine the optimal grouping.Finally,this paper successfully employs a WTA neural network based on DNA strand displacement reactions to accomplish the classification and recognition of printed Chinese numerals.In conclusion,this paper elucidates the principles of constructing a WTA neural network based on DNA strand displacement reactions and demonstrates the application of DNA-based WTA neural networks in the recognition of printed Chinese characters and digits.It also provides valuable references and insights for further research on DNA-based neural networks.