A hybrid spiking neural network model for multivariate data classification and visualization.

• Main Authors: Ming, Leong Yii, Teh, Chee Siong, Chen, Chwen Jen
• Format: Proceeding
• Language: English
• Published: 2011
• Subjects: L Education (General); T Technology (General)
• Online Access: http://ir.unimas.my/id/eprint/9976/1/A%20hybrid.pdf; http://ir.unimas.my/id/eprint/9976/; http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5999509&tag=1

This study proposes a hybrid model of Self-Organizing Map with modified adaptive coordinates (SOM-AC) and Spiking Neural Network (SNN) for multivariate spatial and temporal data visualization and classification. SOM is one of the most prominent unsupervised learning algorithms. Recently, many extensions for SOM have been proposed for temporal processing. However, none of the extensions uses spikes as means of information processing. SNN has potential for qualitative advancements in both biological relevancy and computational power. Therefore, this hybrid learning model is proposed to harness the advantages of both SOM-AC and SNN to produce intuitive multivariate data classification and visualization. Empirical studies of the hybrid model using synthetic and benchmarking datasets yielded promising classification accuracy and intuitive rich visualization.