Greetings from
NeuroDimension!
Makers of NeuroSolutions,
the Neural Network Simulation Environment.
This issue of the
newsletter highlights our upcoming neural network course and includes additional
useful information for users of NeuroDimension products.
In this issue you’ll find:
What’s News?
* Deadline Approaching for Neural Network Course Discount
* NeuroDimension Is Hiring
Designing Neural
Networks
* Symbolic Inputs
* Using genetic algorithm and simplex method to stabilize an oil treatment plant inlet flow
Products and Events
of Interest
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 News?
The deadline for an early registration discount is approaching soon for our November neural network courses in Orlando, Florida. The courses will take place November 13-17, 2000 in Orlando, Florida at the Holiday Inn Hotel & Suites across from the Universal Studios theme parks. The deadline to qualify for a 10% discount is October 9th.
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 textbook, 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.
Our previous courses continue to receive rave reviews from our attendees. In fact, everyone has said that they would recommend our courses to friends. We average a score of 4.6 of 5 when attendees are asked to rate the overall quality of the course.
For details on this new offering, or to sign-up from the Internet, see http://www.nd.com/course/nov_2000.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
NeuroDimension, Inc. is now hiring qualified engineers to work on phase II of our development of a neural network controlled ventilator. Interested parties are encouraged to visit the NeuroDimension web site for more information on this position and previous work on this project.
For information on this position, see http://www.nd.com/employment.htm
For information on phase I of this project, see http://www.nd.com/ventilator/
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
Many times people will want to use inputs that are categorical in nature, not numerical. For instance, the goal may be to predict the price of a house and one of the inputs may be the nature of the lot (e.g. wooded, landscaped, natural, or cleared). Strictly speaking, neural networks can only find relationships between numbers or numeric values, so these categorical values need to be converted to numeric values.
The first reaction is to enumerate the categories (1 = wooded, 2 = landscaped, 3 = natural, 4 = cleared) and use these as inputs. This may provide reasonable results, but the numbering implies a relationship in the data that may not exist. In this case, it implies that a “wooded” lot is more similar to a “landscaped” lot than a “natural” lot. In some cases, this relationship is actually correct. For instance, rating your satisfaction with a realtor (0=unsatisfied, 1 = satisfied, 2 = very satisfied). When there is not necessarily a simple ordered relationship between classes, the best bet is to “unary encode” your inputs. This assigns one new input column to each category and assigns a 1 to that input column when the input is in that category (and a 0 to all other categories). For instance, in the above example, you would create 4 input columns called “wooded?”, “landscaped?”, “natural?”, and “cleared?”. If the lot were landscaped, the 4 inputs would be (0, 1, 0, 0).
NeuroSolutions will do this automatically when you label a column in the Neural Wizard or file component as “symbolic”. The default setting is “unary encoding” and NeuroSolutions will translate that column into multiple columns. One issue to worry about is when you have many different categories, your input space becomes very large (e.g. if you have 100 different categories of property you will create 99 extra inputs to the network). In this case, it may be better to create broader categories. Another option is to use a SOM or Kohonen map to cluster your categories. Once you have trained your SOM and found that similar categories are clustered in similar locations, you can simply use the coordinates of the winning SOM PE as the input to your neural network.
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
Adriano C. Silva
This is a hybrid approach, composed of a genetic algorithm and a linear programming method, to achieve an efficient pipeline network operation. The pipeline network optimization consists of the determination of pump scheduling over a short-term horizon, usually one or more days ahead. The resulting mathematical problem has a dynamic and combinatorial characteristic, in which a sub-optimal solution was obtained through these two mathematical tools in a short computational time. The approach was applied in a Pipeline Network to a study case based on the Patagonia Argentina, which is comprised of 16 tanks and linked pumps, with 66 kilometers of pipelines, that transport the production of more than 100 wells to a pre-processing plant. The goal was to obtain a constant input flow rate at the plant respecting physical and chemical processes requirements.
NOTE: This is just an abstract of the application summary. The entire summary is available at:
http://www.nd.com/application%20summaries/appsum-oilplant.htm
Want to have your solutions spotlighted? We strongly encourage our customers to send their 1-2 page application summaries to submissions@nd.com so that we may post them on our web site at: http://www.nd.com/applicationsum.htm. In each newsletter, we’ll spotlight a new solution and include a link for people to get more information.
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
NeuroDimension supports the 2001 WSES International Conference on Neural Networks and Applications (NNA'01) that will be held in Puerto De La Cruz, Tenerife, Canary Islands -- February 11-15, 2001. Visit http://www.worldses.org/wses/nna or email to nna@worldses.org for more information.
Have an event or product you would like to announce? Contact us at submissions@nd.com.
=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
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
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!