Author: Mengjie Zhang
Source: GZipped PostScript (1020kb); Adobe PDF (1054kb)
Neural networks have been widely applied to data mining since the late 1980s. However, they are often criticised and regarded as a "black box" due to the lack of interpretation ability. This paper describes a domain independent approach to the use of neural networks for mining multiple class objects in large images. During the object mining process, both the classes and locations of the objects are determined. The results suggest that this approach can be used to mine simple and regular objects with translation and limited rotation invariance in large images against a relatively uniform background. The network behaviour is interpreted by analysing the weights in learned networks. Visualisation of these weights not only gives an intuitive way of representing hidden patterns encoded in neural networks for object mining problems, but also shows that neural networks are not just a black box but an expression or a model of hidden patterns extracted/discovered during the data mining process.