[ 
https://issues.apache.org/jira/browse/MAHOUT-383?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12860330#action_12860330
 ] 

Zoran Sevarac commented on MAHOUT-383:
--------------------------------------

Yes, we from Neuroph project are very interested to provide some kind of 
integration. For start maybe you could tell us what do you expect to have from 
the Mahout side, for what and how you are going to use neural networks. Then we 
can consider how that can be accomplished with Neuroph.

Also Jake asked us about some benchmarks regarding performance, accuracy etc. 
and at the moment we dont have that yet. Could you please let us know what 
tests are important to you, and can you provide us some test data or test 
cases. Thanks.

Zoran

> Investigate possibility of integration with Neuroph neural-net library
> ----------------------------------------------------------------------
>
>                 Key: MAHOUT-383
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-383
>             Project: Mahout
>          Issue Type: Task
>          Components: Classification
>    Affects Versions: 0.4
>            Reporter: Jake Mannix
>
> The [Neuroph project  | http://sourceforge.net/projects/neuroph/ ] is an 
> open-source (currently LGPL, but voted to change to ASL to accomodate 
> integration with Mahout) ANN library, complete with this feature-set:
> * Supported neural network architectures
> ** Adaline
> ** Perceptron
> ** Multi Layer Perceptron with Backpropagation
> ** Hopfield network
> ** Bidirectional Associative Memory
> ** Kohonen network
> ** Hebbian network
> ** Maxnet
> ** Competitive network
> ** Instar
> ** Outstar
> ** RBF network
> ** Neuro Fuzzy Reasoner
> *Other features
> ** Small number of the essential base classes (only 10) which can be easily 
> reused
> ** Support for supervised and unsupervised learning rules
> ** An easy-to-follow structure and logic.
> ** GUI tool for neural network development easyNeurons
> ** Image recognition support 
> They are interested in integration, so we should look into what we can do to 
> augment Mahout's capabilities with this.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.

Reply via email to