BUS4650

Business applications of neural networks

K A Smith

6 points
* One 2-hour lecture and one 1-hour tutorial per week
* First semester
* Clayton

Objectives On completion of the subject students should be able to appreciate the advantages and limitations of neural network models for solving a wide range of practical business problems; understand the neural network architectures which are suitable for different types of applications; understand the issues of convergence and training of neural networks; train their own neural networks using commercial software; take a business problem, decide on an architecture, choose training parameters, train the network, test the network and interpret the neural network results.

Synopsis The use of suitable neural network architectures and convergence issues with reference to practical business problems with the use of commercially available neural network software. Applications are taken from several varied areas of business including marketing, business data classification, finance and investment, character recognition, operations/project management and managerial decision making. Real business problems will be solved in tutorials and assignments.

Assessment Assignment 1 (short mid-semester test to assess understanding of the principles and mechanisms in neural network models): 20%
* Assignment 2 (This is a practical assignment designed to assess the student's understanding of the practical issues in solving a real business problem. It involves selection of a neural architecture, resolution of training issues, using commercial software for a particular business application chosen from a provided selection of real world problems): 30%
* Examination (The examination assesses students' understanding of the concepts, theories and application of knowledge and skills attained in the course.): 50%
* Students must pass the examination in order to pass the subject.

Recommended texts

Beale R and Jackson T Neural computing: An introduction IOP, 1994