作者: Ilyes Jenhani , Zied Elouedi
DOI: 10.1007/S10462-012-9360-0
关键词: Computer science 、 System a 、 Immune recognition 、 Classifier (UML) 、 Artificial immune system 、 Machine learning 、 Artificial intelligence
摘要: This paper surveys the major works related to an artificial immune system based classifier that was proposed in 2000s, namely, recognition (AIRS) algorithm. survey has revealed most on AIRS dedicated application of algorithm real-world problems rather than theoretical developments Based this finding, we propose improved version which dub AIRS3. AIRS3 takes into account important parameter ignored by original algorithm, number training antigens represented each memory cell at end learning (numRepAg). Experiments new data sets taken from UCI machine repository have shown taking numRepAg information enhances classification accuracy AIRS.