This part of the IML module will introduce you to neural networks as a pattern recognition tool. Neural networks are a biologically-inspired paradigm for computation. That is, the way in which a result is computed is loosely based on the way in which biological systems react when presented with stimuli. Typically, neural networks are applied to classification problems and function approximation problems. Their defining characteristic is that neural systems "learn" how to perform their computations, and therefore the solution to a problem need not be explicitly programmed - it can be derived from data about the problem. There are many different architectures (structures) for neural networks. The first three lectures will look at the basic principles and some common architectures for neural systems. We will then look at how neural networks have been applied to problems in the bio-informatics area by looking at selected papers from the literature.

It is not my intention to turn you into experts in neural computation - that is not possible in the time we have available. However, you should end up with an appreciation of the way in which neural networks operate, and of the sorts of problems they have been and could be applied to.