Artificial Neural Network Technology : Artificial Neural Network Technology Presented by:Joan Carter Jonathan MacArthurMartin Sand John WellsCharles Shaughnessy Phil Jones The beginning : The beginning Started in the 1940’s
Stunted in the 1960’s
Here to stay in the 1980’s What it does : What it does Uses non-linear statistical data modeling tools
Current computer given step by step instructions
The computer can learn
Optimal solution provided by computer
Robot issues Basic types of applications : Basic types of applications Recognition and identification
Monitoring and control
E-mail spam filtering
Forecasting and prediction
Financial applications What is Neural Networks replacing? : What is Neural Networks replacing? The hype of the artificial intelligence technology had failed to deliver the expected computing ability for these reasons:
The business community increased its collection of data, but the expectation for its usefulness in critical business assessment & decision making was not realized
Failure to extract meaning from complex and imprecise data
Need to replace the conventional approach to define algorithms to solve problem and make predictions when there are no known solution
Sequential batch processing of data is inadequate for the speed at which information is needed Neural in Business Applications : Neural in Business Applications In Medicine
Electronic Nose Business Forecasting & Predictions : Business Forecasting & Predictions Neural Networks in Marketing
Neural Networks in Credit Evaluation
Neural Networks in Forecasting
Neural Networks in Data Mining Other Uses of Neural Applications : Other Uses of Neural Applications Recognition of speakers in communications
Diagnosis of hepatitis
Recovery of telecommunication from faulty software
Interpretation of multimeaning Chinese words
Undersea mine detection
Three dimensional object recognition
Hand written word recognition Benefits : Predictive ability
allows for better control of equipment Example: Pegasus Combustion Optimization Technology Self learning system capable of devising
new solutions for
improved results Self tuning
achieved Less energy used = money saved! Benefits Cost : Pegasus system – NeuSIGHT costs 1/20th the amount to retrofit an existing plant with new hardware. Cost Business Uses for Neural Networks : Business Uses for Neural Networks Recommended business applications for neural networks, based on our research.
Preventing and detecting fraud
Establishing credit worthiness
Predicting default / bankruptcy
Marketing applications Business Uses for Neural Networks : Business Uses for Neural Networks Preventing and detecting fraud
Neural networks can identify patterns of behavior that have led to fraudulent outcomes in the past.
When a similar pattern is repeated, the neural network software flags the event for investigation.
As more patterns are identified and additional fraud outcomes are confirmed, the neural network software will begin to learn these patterns and gain human like intuition. Business Uses for Neural Networks : Business Uses for Neural Networks Establish credit worthiness
Current credit ratings use a small number of predetermined factors to establish an individuals credit worthiness.
Neural networks are able to look for hidden patterns of behavior that may not otherwise be evident to determine credit worthiness.
Again, a neural network has the ability to learn and improve its ability to predict over time. Business Uses for Neural Networks : Business Uses for Neural Networks Predicting default or bankruptcy
Neural networks have the ability to sort through a large amount of data, including customer credit and payment transactions.
Based on this data, the neural networks can look for patterns of behavior that have led to default or bankruptcy outcomes in the past.
As the neural network software continues to scan customer behavior patterns, the neural network can flag a customers account when a concerning pattern of credit or payment behavior emerges, allowing the company to intervene before the customer defaults or files for bankruptcy. Business Uses for Neural Networks : Business Uses for Neural Networks Marketing applications
Neural networks can look at consumer purchase behavior over time to identify purchase patterns.
Tracking customer purchases, the neural network software can predict what the consumer is likely to purchase next.
This will allow marketers using neural networks to target customers who are likely to purchase their products next, giving them a competitive advantage. Vendors : Vendors Microsoft (Excel) - NeuralTools
Calsci - BrainMaker
MATLAB - Neural network Toolbox Product comparison : Product comparison NeuralTools – Easy applications, fast learning and good intergation as most companies already use Microsoft.
MATLAB Neural nework Toolbox requires MATLAB. This is easy to learn and integrates with major operating systems, but it requires additional training.
Calsci – BrainMaker is compatible with Mac and PC, a tailormade system allowing the user to make better and faster decisions. Prices : Prices From free trial versions to $499 and up.
Neural Network Toolbox requires MATLAB, more costly to implement if the company does not already use MATLAB.
Calsci - BrainMaker Reference Material - Wiki : Reference Material - Wiki Available at http://en.wikipedia.org/wiki/Artificial_neural_network
Provides a broad overview of Neural Networks
Numerous links to other reference materials for a better understanding of the various types of neural networks available today Reference Material - YouTube : Reference Material - YouTube Presentation of Neural Networks by Google
-Future of Neural Networks
-Warning: 1 hour long
Lecture about Neural Networks
Prof. Dasgupta, Indian Institue of Technology
Warning: 1 Hour long Reference – Useful Links : Reference – Useful Links Digital Signal Processing : By S. Smith
-Provides a Mathmatical Understanding of Neural Network Structure Possible future implications : For now, only large firms are able to reap the benefits of artificial neural nets
In the future, personal computers may be integrated with artificial neural networks
Could artificial neural networks make unpredictable situations a thing of the past? Possible future implications Business Uses for Neural Networks : Business Uses for Neural Networks In conclusion:
There are many possible business applications for neural networks. Any application that could benefit from tracking past behavior patterns, even hidden behavior patterns, and using them to predict future behavior is a candidate for a neural network. We only explored a handful of possible uses here today.
Thank-you for taking the time to listen to our presentation and we hope you have learned just a little more about neural networks and their possible business applications. References : References Khandelwahl, V. Neural Networks. November, 2007. Presentation. http://www.eusflat2007.cz/category/electronics/
Koprowski, Gene J. Neural-Network Technology Moves into Mainstream. August 2003. http://www.technewsworld.com/story/31280.html
Laboratory of Biomedical Engineering of the School of Electrical and Computer Engineering of National Technical University of Athens
Taylor, James. Live from Business Rules Forum- Intelligent process Automation: the Key to Business Process Automation. October, 2009 http://jtonedm.com/2007/10/24/live-from-business-rules-forum-intelligent-process-automation-the-key-to-business-process-automation/.
Unknown. Directing Intelligence in Business.
Nd.com. NeuroDimension Incorporated. 4 Nov. 2009. < http://www.nd.com>
Cormactech.com. CorMac Technologies Inc. 4 Nov. 2009. http://cormactech.com/neunet/uses.html
Stergiou, Christos and Dimitrios Siganos. “Neural Networks.” SURPRISE 96 Journal vol. 4 (2004). <http://www.doc.ic.ac.uk/~nd/surprise_96/journal/vol4/cs11/report.html>