(Note - I have only seen a demo) This is a collection of modules usable in the program Extend which let you construct and play with neural networks. Extend is a Stella-like program (see review) which lets you graphically construct and solve ODE models. Some of NeuroLabs introductory level modules are things like Adaptive Weighted Summation, Sigmoid, Linear, I/O Control and BP Control. These modules can be used to construct simple neural networks, such as network to do the XOR function, or much more complex networks. Using the Extend interface, you lay out the models nicely, make graphs of the results, and even animate components of the model as it runs. I am not a particular expert in this field and this is rather sophisticated, so you should check the demo yourself if you are interested.

I think these modules would only really be useful in an upper-level modelling course where you were focusing quite a bit of time on neural networks. In that case, having these modules around might be quite nice and a big time-saver. Its also possible that an instructor could use these modules to make a neural network that students could then play with. However, due to the learning curve involved both in learning these modules and in learning how to use Extend, I am not sure this is the best way to introduce neural networks unless you are really going to spend some time on them.

Computer: Macintosh and Windows

Source: Mikuni Berkeley R&D Corp

Demo available on Info_Mac mirrors.

Cost: High

Last update 22nd of January 1996