Classification of tea specimens using novel hybrid artificial intelligence methods

• Published in: Sensors and actuators. B, Chemical Vol. 192; pp. 117 - 125
• Main Authors: Pławiak, Paweł, Maziarz, Wojciech
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.03.2014
• Subjects: Algorithms; Artificial intelligence; Artificial intelligence methods; E-nose; Evolutionary-neural systems; Expert systems; Fuzzy systems; Genetic algorithms; Hybrid systems; Networks; Neural networks; Pattern recognition; Preprocessing; Recurrent neural networks; Tea; Tea; Evolutionary-neural systems; Neural networks; E-nose; Artificial intelligence methods; Hybrid systems; Pattern recognition; Genetic algorithms; Fuzzy systems
• ISSN: 0925-4005; 1873-3077
• DOI: 10.1016/j.snb.2013.10.065

Two innovative systems based on feed-forward and recurrent neural network used for qualitative analysis has been applied to specimens of different fruit tea. Their performance was compared against the conventional methods of artificial intelligence. The proposed systems are a combination of data preprocessing methods, genetic algorithms and Levenberg–Marquardt (LM) algorithm used for learning feed forward and recurrent neural networks. The initial weights and biases of neural networks chosen by the use of genetic algorithms were then tuned with a LM algorithm. The evaluation was made on the basis of accuracy and complexity criteria. The main advantage of the proposed systems is the elimination of the random selection of the network weights and biases resulting in the increased efficiency of the systems.