Encog Machine Learning Framework

Encog 2.4-beta1 has been released. This is a beta release, but it seems to be quite stable. Of course, if you notice any bugs we would like to hear about them. Once no bugs have been reported for a time, we will release an official version. There ... [More] are many new features in this version! It is a fairly large release. Levenberg Marquardt Training, NEAT GPU Acceleration (OpenCL) Many Improvements to Workbench http://code.google.com/p/encog-java/downloads/list http://code.google.com/p/encog-cs/downloads/list [Less]

Wanted to give an update on Encog's use of GPU. It is giving me some very nice results. Right now I am getting about a 10% speed boost on my computer when I engage the GPU. This is with the GPU and CPU working together compared to CPU working alone. Considering I sometimes train for days, this will save a great deal of time. read more

This is as much a question as it is a statement. I am outlining how I see bias neurons being supported in Encog 2.4. Bias neurons have been a very requested feature in Encog for some time now. I want to get this functionality into Encog 2.4. But ... [More] I am looking for some input as to how best to handle this. First of all, Encog already supports something called "threshold values". Threshold values, as they are defined, in Encog support something that I think is very similar to the bias neuron. And I do NOT want to introduce two ways of doing exactly the same thing in Encog. read more [Less]

Encog 2.4 is progressing nicely. Nearly all "new development" is complete. The last frontier of new development is GPU processing. I believe the OpenCL kernals are done. Just need to beef up the support code so that it can use the CPU at the same ... [More] time and divide up big sets better. Also, the OpenCL code is ONLY in the .Net side of things for now. I do not want to translate to Java until I feel the OpenCL code is fairly stable. However, this should be a fast process. I will not have to translate the kernel code at all. Just the interface code. read more [Less]

One of the main points of OpenCL is that I am able to create one OpenCL source and compile it on any platform, either ATI or NVidia. Think of it like SQL, you create a small bit of source code (OpenCL) that your main Java(or C#) application ... [More] executes. So far I've been only using my NVidia Graphics card for Encog testing. Today I tried ATI for the first time. I found that none of my kernals compiled! read more [Less]

Neural networks learn by adjusting numeric values called weights and thresholds.  A weight specifies how strong of a connection exists between two neurons.  A threshold is a value, stored on each neuron that either adds or subtracts from the ... [More] incoming weights from other neurons.  Training is the process by which these weights and thresholds are adjusted to cause the neural network to produce useful results.  read more [Less]

Future versions of Encog will support grid/cloud computing. The upcoming Encog 2.4 release will support limited cloud capabilities, for Encog to communicate status back to the "Encog Cloud". This will allow you to monitor training from anywhere ... [More] , simply by logging into this web site. Encog 2.5 will add full "grid-based" training. You will be able to have more than one computer working on a problem. read more [Less]

You can now purchase the "Programming Neural Networks with Encog 2 in Java" and "Programming Neural Networks with Encog 2 in C#". The ebooks have been out for a little while already. However, the paperback copies can now be purchased as well. The ... [More] following two links contain more information about these two books. http://www.heatonresearch.com/book/programming-neural-networks-encog-jav... http://www.heatonresearch.com/book/programming-neural-networks-encog-cs.... [Less]