An English Letter Recognition Algorithm Based Artificial Immune

Authors:
Chunlin Liang;Lingxi Peng;Yindie Hong;Jing Wang
Affiliations:
Software School, Guangdong Ocean Univ., Zhanjiang, China 524025;Software School, Guangdong Ocean Univ., Zhanjiang, China 524025;Software School, Guangdong Ocean Univ., Zhanjiang, China 524025;Software School, Guangdong Ocean Univ., Zhanjiang, China 524025
Venue:
ISNN 2009 Proceedings of the 6th International Symposium on Neural Networks: Advances in Neural Networks - Part III
Year:
2009

Citing 5
Cited 0

Letter Recognition Using Holland-Style Adaptive Classifiers

Machine Learning
Technical Note: First Nearest Neighbor Classification on Frey and Slate's Letter Recognition Problem

Machine Learning
Coevolving functions in genetic programming

Journal of Systems Architecture: the EUROMICRO Journal - Special issue on evolutionary computing
Artificial immune system for multi-objective design optimization of composite structures

Engineering Applications of Artificial Intelligence
Artificial immune based hybrid GA for QoS based multicast routing in large scale networks (AISMR)

Computer Communications

Quantified Score

Hi-index	0.00

Visualization

Abstract

In the study of letter recognition, the recognition accuracy is impacted by fonts and styles, which is the main bottleneck that the technology is applied. In order to enhance the accuracy, a letter recognition algorithm based artificial immune, referred to as LEBAI, is presented. Inspired by nature immune system, antibody cell (B-cell) population is evolved until the B-cell population is convergent through the learning of each training antigen and the memory cells pool is updated by the optimal B-cell. Finally, recognition is accomplished by memory cells. It is tested by the well-known letter recognition data set of UCI (University of California at Irvine). Compared with HSAC (Letter Recognition Using Holland-Style Adaptive Classifiers), LEBAI showed that recognition accuracy is increased from 82.7% to 95.58%. LEBAI achieves the same recognition accuracy for the letters of different fonts and styles, or stretched and distorted randomly.