Title :
Research on handwritten numeral recognition method based on improved genetic algorithm and neural network
Author :
Yan, Tai-Shan ; Tao, Yong-qing ; Cui, Du-wu
Author_Institution :
Hunan Inst. of Technol., Hunan
Abstract :
Considering the limitation such as premature convergence and low local convergence speed of genetic algorithm, some improvements were made for classical genetic algorithm. Firstly, a help operator was used to help individuals of population according to the given probability. Secondly, the genetic individuals were separated into male individuals and female individuals, and consanguinity was fused into individuals. Two individuals with different sex could reproduce the next generation only if they were distant consanguinity individuals. Based on this improved genetic algorithm, an evolved neural network algorithm named IGA-BP algorithm was proposed. In this algorithm, genetic algorithm was used to optimize and design the structure, the initial weights and thresholds, the training ratio and momentum factor of neural network roundly. The disadvantage of neural networks that their structure and parameters were decided stochastically or by one´s experience was overcome in this way and the surge of algorithm was restrained. IGA-BP algorithm was used to recognize handwritten numerals, a recognition model of handwritten numerals based on BP neural network was found, and the handwritten numeral recognition scheme based on IGA-BP algorithm was proposed. The experimental results show that this algorithm is better than SGA-BP algorithm and traditional BP algorithm in both speed and precision of convergence, We can obtain a better recognition effect using this algorithm.
Keywords :
backpropagation; convergence; genetic algorithms; handwritten character recognition; neural nets; probability; stochastic processes; convergence speed; handwritten numeral recognition method; help operator; improved genetic algorithm-backpropagation algorithm; neural network; probability; Algorithm design and analysis; Character recognition; Convergence; Genetic algorithms; Handwriting recognition; Neural networks; Pattern analysis; Pattern recognition; Surges; Wavelet analysis; BP algorithm; Genetic algorithm; Handwritten numeral recognition; Neural network;
Conference_Titel :
Wavelet Analysis and Pattern Recognition, 2007. ICWAPR '07. International Conference on
Conference_Location :
Beijing
Print_ISBN :
978-1-4244-1065-1
Electronic_ISBN :
978-1-4244-1066-8
DOI :
10.1109/ICWAPR.2007.4421630