Neural Networks
for Business Applications

Introduction:

• Emerging technologies in decision/information sciences.
• Expert systems, fuzzy methodologies, or neural networks?
• Overview of neural networks.
• Major components of neural networks.
• What can neural networks be used for?

Selected Neural Networks:

• Taxonomy of neural networks.
• Backpropagation net and its variations.
• Counterpropagation nets.
• Learning vector quantization (LVQ) net and its variations.
• Self-organizing map (SOM) / Competitive learning net.
• Interactive activation and competitive learning net.
• Adaptive resonance theory (ART) net and its variations.
• Hopfield nets / Simulated annealing net.

System Development Issues:

• System development process.
• How to represent data?
• Selection of neural network paradigm (topology).
• Determination on the number of hidden layers and network size.
• Selection of propagation/activation/transfer functions.
• Selection of learning rule and parameters.
• How to initialize the training weights.
• Preparation and determination on the training/testing data sets.
• Selection of development tool(s).
• System integration and maintenance issues

Illustration/Demonstration of Applications:

• Forecasting method selection/time series analysis.
• Cell formation for cellular manufacturing.
• Bank bankruptcy prediction.
• Customer credit / loan evaluation.
• Corporate bond rating and prediction.
• Quality control chart analysis and interpretation.

Implementation and Future Directions:

• Strengths and weaknesses of neural networks.
• Summary of potential applications.
• Strategies for successful implementation.
• Integration with expert systems/fuzzy methodologies/genetic algorithm.

Back to Course Page