MAKHFI.COM
Fascinating World of Neural Nets  
Neural Network Forums
Home PageHome Page : General Neural Networks Topics : Reliability of ANN's
  You are currently not logged in. You can view the forums, but cannot post messages. | HOME | Log In | Register | Search | Help
Post a Reply on This Topic Post a Reply on This Topic

Author Topic: Reliability of ANN's
trilogy Posted: 10-Apr-11 18:09
  Edit Edit
 
Email the Author Mail   View Author's Profile Profile  
Are neural networks safe for businesses to use? No one can really explain how these systems make their predictions yet they seem to be effective; but how can a business owner like myself rely on their predictions? Given that I have no assurances of their reliability what possible consequences might a business owner face in using neural networks across a broad range of applications? Also I hear alot about back-propagation vs. self-organizing but if you dont initially provide some training how does a self-org system train itself or even know where to look for patterns or commonalities?

Curious to learn more and hoping for some input, thanks.
 
pejman Posted: 11-Apr-11 15:55
Delete Delete    Edit Edit
 
Email the Author Mail   View Author's Profile Profile  

Trilogy,

Well, that is actually one the common criticism of neural networks.
I would say the answer depends on your application.

Many businesses are already using neural nets, consciously or unconsciously.
Think of credit scoring or fraud detection at bank!

As a business user/owner focus on providing clear application requirements.
Your IT person should find out if ANN is a good fit or not.


As for self-organized systems, there is still a need for training. The difference is that training data is not labeled (classified).

The training algorithm applied dimension reduction to reduce the number of random variables under consideration (training data). This can be used for feature extraction or clustering.

Hope that helps.

-Pejman
 

Post a Reply on This Topic Post a Reply on This Topic
 

Copyright © 2001-2003 Pejman Makhfi. All rights Reserved.