Application Of PSO-RBF Neural Network In DNA Sequence Classification

Radial Basis Function(RBF)neural network is a typical feedforward network.It has only a simple structure of the hidden layer,so it is easy to learn,not easy to minimize locally,and has strong generalization ability,which shows its superiority in many fields.In this paper,the main research object is PSO-RBF neural network algorithm.In terms of PSO algorithm improvement,the convergence inertia weight is selected to replace the fixed value.The performance of the improved algorithm is tested by the test function,and the results show that the algorithm has stronger local and global convergence ability.The PSO-RBF neural network algorithm is applied to the DNA sequence classification problem for the first time,and a DNA sequence classification model based on PSO-RBF neural network is constructed.DNA sequence classification is an important part of bioinformatics.The study aims to predict the class of unknown DNA sequences to understand their characteristics,which is important for judging whether they belong to hidden species,alien species or endangered species.Feature extraction is an important part of the non-alignment method of DNA sequence classification.This paper proposes a new feature extraction method based on the classical k-mers method: firstly,the base transfer probability is used to replace the single base and double base frequencies when k=1 and k=2,and then the amino acid type is used instead of three-base frequency when k=3,and finally the principal component analysis method is used to reduce the dimension of the feature vector.Compared with the k-mers method,the DNA sequence feature vector extracted by this method not only has a small data dimension,but also has stronger biological significance.Finally,the model was tested using the real DNA sequence of the National Center for Biotechnology Information(NCBI).The test results show that the correct classification rate of the DNA sequence is 94.90%,which has high practicability.