Classification of time series data: a synergistic neural networks approach

Application of neural networks have been plentiful in past decades. One of the development in recent years is the synergy of neural networks. There have been reported in literature under different terms and scopes, there has yet to be an overview study of such approach. Even though classification is probably most prolific application area of neural networks, relatively few have been carried out on times series data. One such application is in classification of time series patterns which are commonly occurred, especially in real time control system. This paper presents an overview of synergistic approach to neural networks. It systematically classifies the approach according to 2 facets. It follows by presenting how a synergy of neural networks could improve the performance in classification of time series control chart patterns when signals are highly noisy. This work illustrates the benefit of synergistic approach in neural networks and points out the importance of selecting appropriate neural network architecture for such classification.