Greetings
from NeuroDimension!
The World Leader in Neural Network Software
This issue of the NeuroDimension
newsletter highlights new releases of NeuroSolutions and its associated
products and the upcoming neural network course, along with tips and techniques
for using NeuroSolutions more effectively.
In this issue you’ll find:
* NeuroSolutions v4.13 Now Available
What’s New
and News?
* Neural Network Course
Completed
* New Neural Network Course
Added
* New TradingSolutions Release
Available
* Discriminating Between
Classes with Different Frequencies
* Neuro-Fuzzy Architecture
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.
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
NeuroSolutions v4.13 is now
available from NeuroDimension. This release addresses a handful of minor bug fixes,
mostly related to the new features such as, Neuro-fuzzy, SVMs and Genetic
Optimization. There have also been a few new features added since v4.0
including:
* NeuroSolutions for Excel
support of Office XP
* Support for reading 24-bit
bitmaps
* Enhanced documentation for
some of the new components and algorithms
* Enhanced performance for
the Neuro-Fuzzy networks
If
you already have NeuroSolutions v4.0 or v4.1 installed on your computer you can
upgrade to v4.13 by downloading and running the patch from: http://www.nd.com/support/ns_patch.htm
The
complete installation program is available for download at: http://www.nd.com/download.htm
Note that if your evaluation copy of NeuroSolutions v4.0 has
already expired, upgrading to v4.13 will extend your evaluation period by
another 60 days.
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
We successfully completed another neural network and
NeuroSolutions course in early May. It included attendees from Saudi Arabia,
Denmark, Great Britain, and all over the US. We always enjoy getting to know
our customers better, especially when it gives us a chance to discuss their
applications. Once again, we received rave reviews from our attendees, giving
us an average score of 4.6 of 5 when asked to rate the overall quality of the
course.
The next neural network course has been scheduled for
November 5-9, 2001 at the Grosvenor Resort, located in the Walt Disney World
Resort in Orlando, Florida. A 10% discount is available for early registration,
so be sure to register today!
Our course format allows both novice and advanced users to find
a suitable course. Offered courses include: "Introduction to
NeuroSolutions", "Fundamentals of Neural Networks and
NeuroSolutions", and "Advanced NeuroSolutions". In addition,
each of the courses includes updates on the new features of NeuroSolutions 4.0,
geared towards each level of user.
The courses include a copy of our interactive book, Neural
and Adaptive Systems: Fundamentals Through Simulations. We are also happy
to work with attendees who would like to use their own data in the sample
projects.
For details on this new offering, or to sign-up from the
Internet, see http://www.nd.com/course/nov_2001.htm
For general ND course information, see http://www.nd.com/course
For more information and samples of the interactive book,
see http://www.nd.com/products/nsbook.htm
TradingSolutions v1.21 Build 010614 is
now available for download. As with previous releases, version 1.21 helps you
make better trading decisions by combining traditional technical analysis with
state-of-the-art artificial intelligence technologies. Version 1.2 added
easier-to-follow step-by-step tutorials, sample models, and new signal analysis
options. Version 1.21 fixes minor bugs and adds many new built-in functions.
For more information about
TradingSolutions and downloading a FREE evaluation copy, see http://www.tradingsolutions.com
To stay informed with the latest
TradingSolutions updates and tips, be sure to sign up for the TradingSolutions
newsletter.
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
Often times a classification problem may have a large number
of examples of one class and only a few in another. For instance, in medical
classification there may be 90 exemplars of data for patients who do not have a
rare disease and only 10 exemplars for patients who do. Trying to classify this
data can be very difficult for a neural network. The reason is that the network
does not have enough information (e.g. exemplars) about the sick patients
relative to the information about the healthy patients. When this happens the
network has a difficult time detecting the smaller class and typically makes
the simple decision – that every patient is healthy.
The easiest way to deal with this problem is to give more
emphasis to the smaller class. For instance, if you have 90 healthy patients
and 10 sick patients, the errors or gradients from the sick patients will be
weighted 9 times more than the ones from the healthy patients. This is
equivalent to duplicating each of the sick patients 9 times in your data set,
thus making the number of exemplars in each class the same. Whenever you have a situation where the
network is focusing too much on the larger class(es), you should “weight the
gradients”.
Weighting the gradients has been a feature in NeuroSolutions
for quite some time, but does not get much attention. To turn this feature on,
open the properties of the “BackStaticController” and click on the “Weighting”
page. Click the check box that says “weight the gradients” and then click the
“Assign weights to file” button. This will automatically read the desired file
and calculate the frequency of each class. It will then scale the gradients
from each exemplar so that each class will get equal gradient weighting. If you
desire, you may create your own “weighting file” and instruct NeuroSolutions to
use it via the “Weighting File” button.
NeuroSolutions for Excel v4.1 now includes a check box on
the training panel labeled “For classification problems, make classes evenly
weighted” that will automatically set NeuroSolutions to weight the gradients.
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
Fuzzy logic is a superset of conventional (Boolean) logic
that has been extended to handle the concept of partial truth -- truth values
between "completely true" and "completely false". It can
often simplify a problem by using a priori
information to divide the input space. This knowledge is encoded into
membership functions, which are used to determine the decision boundaries.
A neuro-fuzzy system is a neural network / fuzzy
logic hybrid that combines the best of both worlds. It uses fuzzy logic to
interpret the inputs and backpropagation to map the inputs to the desired
output. In addition, backpropagation can be used to adapt the parameters of the
fuzzy membership functions so that a priori knowledge of the inputs is
not required.
The NeuralBuilder of NeuroSolutions
4 includes a neural model called the coactive neuro-fuzzy inference system, or
CANFIS for short. A good introduction to this architecture is included with the
NeuroSolutions demos (under the Help menu) and within the NeuralBuilder
documentation. A more detailed description can be found in Neuro-Fuzzy and
Soft Computing by J.-S. R. Jang, C.-T. Sun, and E. Mizutani
(Prentice Hall, 1997).
The CANFIS model can be useful for
those problems that have non-discrete or poorly defined input data. Due to the
exponential relationship between the number of input columns and the number of
internal processing elements, it is recommended that this model be limited to
data sets that have very few (five or less) inputs.
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
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: feedback@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.nd.com/mailinglist.htm
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!