[jira] [Commented] (HORN-10) Implement convolutional neural network based on Neuron-centric

["Edward J. Yoon (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HORN-10?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15395178#comment-15395178 ] 

Edward J. Yoon commented on HORN-10:
------------------------------------

I've wrote some code here https://github.com/edwardyoon/incubator-horn

!http://neuralnetworksanddeeplearning.com/images/simple_conv.png!

We can define the layers like below:
{code}
    job.inputLayer(features);
    job.addLayer(24 * 24, Sigmoid.class, ConvolutionNeuron.class);
    job.addLayer(12 * 12, Maxout.class, StandardNeuron.class);
    job.addLayer(100, Sigmoid.class, StandardNeuron.class);
    job.outputLayer(labels, SoftMax.class, StandardNeuron.class);
{code}

In this case, the number of parameters is 5*5*20 (50 feature maps), the number of neurons of conv layer is 24 * 24. Each neuron will receives the region of the image and set of kernel filters, and computes the output. We choose 1) strategy.

1) neuron receives the region of the image and set of kernel filters. 





> Implement convolutional neural network based on Neuron-centric
> --------------------------------------------------------------
>
>                 Key: HORN-10
>                 URL: https://issues.apache.org/jira/browse/HORN-10
>             Project: Apache Horn
>          Issue Type: New Feature
>            Reporter: Edward J. Yoon
>            Assignee: Zachary Jaffee
>




--
This message was sent by Atlassian JIRA
(v6.3.4#6332)