How to save input-vector to the output layer of Neural Network

[ankit rawat]

Hi

I have implemented and trained the LeNet 5 Convolutional Neural Network.

Now while testing new samples, for each sample, I am interested in saving the vector output from the penultimate layer of the Neural Network i.e. "input vector to the output layer". 

Kindly let me know how I can accomplish this. Do I need to make some change in the source code or is there any way through adding/changing some parameter in the configuration file.

Thanks and regards

Ankit

[the_minion]

https://groups.google.com/forum/?fromgroups#!topic/eblearn/fMq7bbCglJ0

[ankit rawat]

Hi

Thanks for the reply. Sorry for the delay in reply.

I went through the post, but as you have mentioned in the post the dump file contains the output vector of only the last layer (whereas I am looking for the output vector of the previous layer). For easy reference I am quoting you from the post:  

"Oh nevermind, it does look like the output is the last layer's output (before it determines what class it is)."

Also I get intermittent crashes/segmentation faults while running the MNIST Digit classification demo. Has it something to do with my system configuration?

My system config: 

================================

OSX  10.8.1

Processor  2.5 GHz Intel Core i5

Memory  4 GB 1600 MHz DDR3

Graphics  Intel HD Graphics 4000 384 MB

================================

Following is the error log snippet:

================================================

Saving outputs to .//detections/

Wrote configuration to .//detections/mnist.conf

Not using Intel IPP.

Initializing 1 detection threads.

initializing thread (unsynched outputs)

Saving outputs to .//detections/

Initializing directory camera from: /Users/ankitrawat8/Documents/Handwriting_Recognition/eblearn/eblearn/tools/data/mnist/

Image search pattern: .*[.](png|jpg|jpeg|PNG|JPG|JPEG|bmp|BMP|ppm|PPM|pnm|PNM|pgm|PGM|gif|GIF|mat|MAT) in /Users/ankitrawat8/Documents/Handwriting_Recognition/eblearn/eblearn/tools/data/mnist/

Segmentation fault: 11

=================================================

Thanks in advance.

Regards

Ankit

--
You received this message because you are subscribed to a topic in the Google Groups "eblearn" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/eblearn/HcQrblE05aM/unsubscribe?hl=en.
To unsubscribe from this group and all its topics, send an email to eblearn+u...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.