Author(s)

You Zhou

Corresponding Author

You Zhou

ABSTRACT

Neural network is widely used in financial analysis. However, there are many indicators to describe financial ratio. If these indicators are used as model inputs, the network structure will be too large. Independent component analysis (ICA) is a new signal processing technology developed in recent years, which decomposes multi-channel observation signals into several independent components by optimization algorithm according to the principle of statistical independence. In this paper, we propose to use ICA to reduce the dimension of high-dimensional indexes in the small sample problems such as financial crisis analysis, reduce the network scale and improve the reliability of analysis and prediction on the premise of retaining the vast majority of sample information.

KEYWORDS

Crisis analysis, neural network, independent component analysis, dimension reduction