Applications of Neural Networks

Can you use neural networks in your business? Here are some past applications of neural networks (or similar statistical programs). Some examples are taken from our customers, and others from articles and product literature of several neural network products.

Neural Nets have been used to:

• Predict staffing requirements at different times of the year and different conditions. Brooklyn Union Gas Corp predicts in advance the number of crew members who will be needed for service calls based on the time of year, predicted temperature, and day of the week.
• Predict which job a job applicant is best suited for. (Brooklyn Union Gas)
• Predict which customers will pay their bills (Brooklyn Union Gas)
• Spot odd trading patterns (this is how Ivan Boesky, the rogue trader, was caught).
• Predict the properties of chemical mixtures.
• Diagnose diseases (One of our customers trained a net that outdid an expert system in diagnosing smell disorders)
• Predict the stock market, the futures markets, etc.
• Flag faulty parts on an assembly line.
• Regulate industrial processes using inputs from sensors at different points in the process.
• Classify medical ailments (such as hearing losses) and classify living things, and classify cells as cancerous/non-cancerous.
• Predict pollution based on the composition of trash coming into an incinerator.
• Predict Sales
• Predict Costs
• Predict a company's corporate bond rating
• Appraise Real Estate
• Predict the outcome of sports events (such as horse racing).
• Predict Solar Flares
• Predict the length of survival for medical patients with ailments such as cirrhosis of the liver.
• Recognize welds which are most likely to fail under stress
• Test beer: Anheuser-Busch: Identifies the organic contents of its competitors beer vapors with 96% accuracy.
• Predict which prison inmates could benefit from less expensive alternative programs. (Delaware correctional system)

There are many more examples than can be listed here. In general, a neural net can capture the hidden relationships in historical data, which allows you then to predict future trends in that data.