基于进化神经网络的RNA二级结构预测方法

本文提出了一种基于进化神经网络预测RNA二级结构的方法.该方法利用进化算法优化了神经网络结构和权值,改善了传统BP人工神经网络容易陷入局部最优等缺陷.实验结果表明该算法能够获得一个最优结构和权值的神经网络,利用该网络去预测不带伪结的RNA二级结构能获得理想的效果,敏感性和相关系数都有所提升,证明了该改进算法的可行性.

An evolutionary neural network approach to predict RNA secondary structure

A novel RNA secondary structure prediction approach based on optimized Neural Network (NN) is proposed. The structure and the weight of NN are optimized by evolutionary algorithm. Through optimizing the structure and weight, the deficiency such as trapping into the local area of traditional NN is improved.Experiments show that this algorithm can achieve an optimal structure and weight of the NN. Using the NN to predict RNA secondary structure can obtain ideal effect. And the sensitivity and correlation coefficient of RNA secondary structure are improved. Simulation validates feasibility of the proposed approach.