Greetings
from NeuroDimension!
The World Leader in Neural Network Software
This issue of the NeuroDimension
newsletter highlights news about our neural network courses and two new product
upgrades, along with a helpful article on selecting neural network inputs.
In this issue you’ll find:
What’s New
and News?
* Neural Network Course Completed
* New Neural Network Course
Added
Product
Upgrade Announcements
* TradingSolutions v2.0 Now
Available
* NeuroSolutions v4.18 Upgrade
Available
* Selecting Inputs
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.
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
We successfully completed another course on neural networks
and NeuroSolutions earlier this month. Attendance was down a little due to the
recent decline in air travel, but those who attended seemed very pleased that
the instructors were able to spend the extra time to address their individual
problems and questions. In fact, we received the highest marks yet from the
surveys given to the attendees – we received an average score of 4.67 of 5 when
asked to rate the overall quality of the course.
The next neural network course has been scheduled for April
29 - May 3, 2002 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". The courses
include a copy of our interactive book, Neural and Adaptive Systems:
Fundamentals Through Simulations.
For details on this new offering, or to sign-up from the
Internet, see http://www.nd.com/course/may_2002.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 v2.0 is now available from NeuroDimension. TradingSolutions
is software that helps you make better trading decisions by combining
traditional technical analysis with state-of-the-art artificial intelligence
technologies. Version 2.0 introduces many new features and enhancements, including:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
- - - - - - - - - - - - - - - - - - - - -
The TradingSolutions Solution Service
The TradingSolutions Solution Service is a service
maintained by NeuroDimension (the creators of TradingSolutions). It provides
you with models, data, and commentary for a select group of stocks that you can
follow on a daily basis. These models also provide a valuable learning tool as
examples for creating your own models.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
- - - - - - - - - - - - - - - - - - - - -
Easier-to-Follow Signals
New features make it easier to track and trade signals with
TradingSolutions. Easily identify the signals you want to trade. Add those
signals to your portfolio along with custom analyses of how they would have
traded. Even keep notes about your successes right in your portfolio!
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
- - - - - - - - - - - - - - - - - - - - -
Improved Demonstrations and Tutorials
The animated demos in TradingSolutions have been completely
rewritten to tell you exactly what TradingSolutions can do for you, as well as
how you can start using TradingSolutions for your trading today.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
- - - - - - - - - - - - - - - - - - - - -
You can find out more about
TradingSolutions and download a free evaluation version from our newly
redesigned TradingSolutions website at http://www.tradingsolutions.com
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
NeuroSolutions v4.18 is now available from
NeuroDimension. This release addresses a few minor bugs outlined at http://www.nd.com/support/nsimprove.htm
Please
report any other bugs that you encounter to support@nd.com
so that they will be fixed for the next upgrade patch.
If you already have NeuroSolutions v4.0 or v4.1 installed on
your computer you can upgrade to v4.18 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
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
Frequently
users have a problem of trying to decide which and how many of the myriad of
possible inputs should be used to solve the problem at hand. For instance, if
the problem is predicting the price of oil tomorrow, should we include the
price of other commodities in the input, and if so, which ones?
The first
reaction of many people is to use all of the possible inputs and assume the
network will find the right ones. This is highly inefficient and more
importantly causes severe generalization problems. For small neural networks, a
good rule of thumb is that if you have N input columns into your network, you
would like to have at least 40*N exemplars (e.g. samples or rows) of training
data. This is a simplification of the more general 10*number_of_weights rule we
have used in the past.
The most
important (and obvious) method for selecting which inputs to use is intuition
and reason. If you think the input is related to the output, then use the
data. If there is reason to believe
that it is not, then don’t include it.
Intuition, whether you know it or not, is always the first thing you use
in selecting the subset of data you will even consider using.
The next
methodology you should consider is correlation between input and output (this
can be done in NeuroSolutions for Excel, TradingSolutions, or external
statistical packages). Correlation measures the relationship between two
variables and ranges from -1 to 1. Any correlation between input and output
that is significantly different from 0 is probably useful in the neural
network. The important thing to remember is that correlation measures linear
relationships, so if a relationship only exists for a certain range of inputs,
correlation may not adequately reflect how useful that input may be to a
nonlinear system.
Another
technique that is useful is sensitivity analysis. This was described in our June
2000 newsletter and measures how important each input in the network is in
making the decision/prediction.
Another
option is to use genetic algorithms to selectively turn inputs on and off in
order to find the optimal set of inputs to solve the current problem. Genetic
optimization can be used in NeuroSolutions (NeuralBuilder and NeuralExpert),
NeuroSolutions for Excel (train genetic), and TradingSolutions.
Our
recommendation to get from the potentially infinite possibilities to a
reasonable set is as follows:
1. Use your intuition as a first guess as to
what is reasonable. Narrow your choices down to no more than 100 or so.
2. Use correlation analysis to further narrow
your choices to no more than 20.
3. Use genetic optimization to find the best
subset of those 20.
4. Use sensitivity analysis to ensure that all
of that subset is actually important in determining the appropriate result.
Finally,
remember that the number of inputs you select should be supported by the basic
rules of thumb for the number of samples of data you have available.
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
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!