Benchmarking and Comparing Encog, Neuroph and JOONE Neural Networks

Submitted by Marszy on Tue, 06/08/2010 - 05:19

Using Encog in Java

"As you can see from my two articles, the clear winner is Encog. It provides a clean and easy to use API and stunning performance. The performance of Encog currently cannot be matched."

http://www.codeproject.com/KB/recipes/xor-encog-neuroph-joone.aspx
http://www.codeproject.com/KB/recipes/benchmark-neuroph-encog.aspx

Nice work! :)

Best

/Mårten

Submitted by SeemaSingh on Tue, 06/08/2010 - 06:23.

I think JOONE is pretty much an orphaned project at this point. I was a bit shocked to see that JOONE beats Neuroph.

Submitted by cogpsych on Tue, 06/08/2010 - 06:29.

I had a vision based neural network that was running under Neuroph. It was taking over a week to train, as I added more examples to it. Encog does this in a matter of hours. It was actually an easy translation from Neuroph to Encog.

Thanks guys, I MIGHT actually get my research done before fall semester. Using Neuroph I don't think there were enough "computer cycles" during the summer to do it!

I think Encog has very much raised the bar for neural network frameworks. Really advanced features such as all sorts of training, multi-core and even GPU support.

I only wish I had seen that article sooner, would have saved me a ton of time.