A New Approach for Pattern Recognition with Neuro-Genetic System Using Microbial Genetic Algorithm

Artificial Intelligence (AI) has been used extensively to solve different types of pattern recognition problems such as disease diagnosis and classification problems. Due to poor performance in recognition of patterns, several methods have been used. Among them genetic algorithm (GA) have shown better performance than other algorithms. Microbial Genetic Algorithm (MGA) or Bacterial Genetic Algorithm is one of the newer branch of GA. MGA follows the evaluation procedure of microbial which gives better results in pattern recognition. In this paper, microbial genetic algorithm with neural network approach for fitness calculation has been developed and it is used for performance analysis of different pattern recognition problems. Proposed algorithm is named as Microbial Neuro-Genetic Algorithm (MNGA). Advantages of MNGA over simple genetic algorithm (SGA) have also been discussed. XOR, Breast Cancer, Diabetes, Heart Diseases, Glass and Card classification problems are taken from UCI machine learning repository dataset as sample problems for performance analysis which shows that this method provides good performance for different types of problems and thus reduces the need for different types of methods for different types of problems.