This issue of the NeuroDimension newsletter highlights the latest developments at NeuroDimension, as well as providing insights on designing effective neural network models.
In this issue you’ll find:
What’s New and News?
* Neural Network Course Discount Ends Soon!
Product Upgrade Announcement
* NeuroSolutions v4.21 Now Available
Designing Neural Networks
* Training with Multiple Outputs
Note: You are receiving this newsletter because you requested to stay informed concerning new developments at NeuroDimension. If you would like to stop receiving these newsletters, please see the bottom of this newsletter for removal instructions.
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
What’s New and News?
Neural Network Course Discount Ends Soon
The next neural network course has been scheduled for April 28 - May 2, 2002 at the Doubletree Hotel and Conference Center located in Gainesville, Florida. A 10% discount is available for early registration before March 21, so be sure to register today!
Our course format allows both novice and advanced users to find a course suitable to their background and interest. Offered courses include: "Introduction to NeuroSolutions", "Fundamentals of Neural Networks and NeuroSolutions", and "Advanced NeuroSolutions". The courses include a copy of our interactive book, Neural and Adaptive Systems: Fundamentals Through Simulations.
For details on this new course offering, or to sign-up from the Internet, see http://www.neurosolutions.com/products/course/april_2003.html
For general ND course information, see http://www.neurosolutions.com/products/course/
For more information and samples of the interactive book, see http://www.neurosolutions.com/products/nsbook/
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
Product Upgrade Announcement
NeuroSolutions v4.21 Now Available
NeuroSolutions v4.21 is now available from NeuroDimension. This release addresses a few minor bugs and usability improvements. A complete list of fixes and improvements can be found at http://www.neurosolutions.com/downloads/nsimprove.html.
If you already have NeuroSolutions v4.0 or higher installed on your computer, you can upgrade to v4.21 by downloading and running the patch from http://www.neurosolutions.com/downloads/patches.html.
The complete installation program is available for download from http://www.neurosolutions.com/download.html.
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
Designing Neural Networks
This Issue: Training with Multiple Outputs
When you have a problem that has multiple outputs, a common question is whether you should create a single network with multiple outputs or multiple networks with a single output each. Like many questions, the answer is that it depends.
From a "power to model" standpoint, multiple networks each with a single output allow more flexibility and thus potentially better results for each model. For instance, each single-output model can use different inputs and have different hidden layer weights (i.e. different intermediate transformations/representations of the data). In situations where the outputs are not very related and the necessary inputs are likely to be different for each output, multiple single-output networks are likely to produce much better results.
On the other hand, if the outputs are related and "solving" or modeling one output may use much of the same information as the others, then using a single network with multiple outputs is likely to provide better results. The reason for this is that each different output (e.g. desired signal) is providing more information to the network to help it train better. This is particularly beneficial when you do not have a large number of exemplars, which leads to generalization problems. The multiple outputs force the hidden layers to create more general relationships that can simultaneously solve all of the problems, instead of overtraining on any one particular problem.
Some researchers even create new desired outputs for this very reason, to create a system with better generalization. For example, if you are predicting an exchange rate between the Euro and the US Dollar, you can add a second output which is the exchange rate between the US Dollar and the Euro. It has been shown that this can improve the performance of the network.
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
Comments or Suggestions?
We appreciate your feedback! Please send us your comments or suggestions concerning this newsletter, our web site, or part of the NeuroDimension product line. Write to us at: newsletters@nd.com
Have questions about NeuroDimension products or training services? Send your questions to: info@nd.com
Some mail readers may display an attachment at the bottom of this newsletter. This is typically caused when the mail reader is set to display text only. The original, formatted version is sometimes included as an attachment.
This issue and previous issues of this newsletter are available on the NeuroDimension web site at: http://www.neurosolutions.com/newsletters.html
If you would prefer not to receive these newsletters or subsequent product updates from NeuroDimension, please reply to this letter with the subject heading changed to the word REMOVE.
Thank you again for your support of NeuroDimension products!