Skip to content
A fast, robust online deep learning architecture based on DeSTIN.
C C++ Shell
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
Python
include
resources
src
CMakeLists.txt
LICENSE
README.md
swig_destina_common.i
update-all.sh

README.md

Destina

##Introduction##

A DeSTIN implementation.

This project is developed under Ubuntu 12.04.3 and Mac OS X.

##To-do List##

  1. Information filter based sparse coding.
  2. Fast inverse (optional)
  3. Test on Gabor wavelet dictionary and natural image with sparse coding. [DONE 20140422]
  4. Clustering algorithm for upper layer. [DONE changed to sparse autoencoder]
  5. Test on second layer learned representation.
  6. Centroids/Dictionary visualisation.
  7. Signal composition and decomposition [DONE 20140425].
  8. General mapping between layers [DONE 20140504].
  9. Performance evaluation.
  10. Memory leaking. [DONE 20140622]
  11. Interface for CIFAR image [DONE 20140621].
  12. Some visualisation tools for image dataset. [DONE]
  13. Experiment result saving functions (needed to rescale to uchar space).
  14. Experiment with CIFAR images [DONE 20140622 (UPDATED IN 20140729)].

##Updates##

  1. Gabor wavelet dictionary design [2014-04-22]
  2. Image whitening updated [2014-04-24]
  3. Information filter based sparse coding and test updated [2014-04-25 (undone)]
  4. Main components of DeSTIN network are sketched [2014-04-27]
  5. Sparse Autoencoder is updated [2014-06-14]
  6. General Refinement of DeSTIN architecture is updated [2014-06-21]
  7. CIFAR classification example sketch [2014-06-22]
  8. Revised contrast normalisation and whitening [2014-06-23]
  9. CIFAR classification on complete set [2014-07-29]
  10. Nao action learning update (up to 9 scenes) [2014-08-04]
  11. Destina Python support by SWIG [2014-08-09]

##Setup##

  1. Clone the whole project (Matlab codes also provide some insight of the project)

  2. cd in destina folder and configure with cmake

    $ mkdir build
    $ cd build
    $ cmake ..
    $ make
    
  3. Some demonstration programs can be found in src, and executable can be found in build/bin.

  4. If you want to update the program (fork and clone from your own account), you can simply use update-all.sh, run following code in terminal:

    $ sh update-all.sh
    
  5. How to set up CIFAR-10 test: download CIFAR-10 dataset and unzip it under destina folder. Then after you build the project, under build/bin you should run classify, then if nothing wrong, your test should go well. All the details are specified in src/classify_test.cpp. You can change line 20 to specify which dataset you are using. Currently, I just wrote up a example on one training dataset, you can do more on the rest by modify the program. To ensure your are on the right track, here is the folder structure from my project

    /destina
        /build
        /cifar-10-batches-bin
        CMakeLists.txt
        /include
        LICENSE
        README.md
        /resources
        /src
        /Python
        /swig_destina_common.i
        update-all.sh
    

##Notes##

  1. Example of Gabor Wavelet dictionary

    Gabor Wavelet Dictionary

  2. Example of Whitened Image

    Whitened Image

  3. In C++ implementation, can only learn some representations roughly. However the performance is not acceptable, still need to figure out it's dictionary's problem or implementation's problem. Furthermore, the speed is not acceptable for now.

  4. Example of reconstruction image with sparse autoencoder

    Sparse Autoencoder Reconstruction

  5. In this implementation, to ensure that DeSTIN is started from a better place, I employed a bottom-up batch training to train the autoencoder in each layer with a batch of images. And after this pre-training, the training stage will be performed as usual and the speed is quite fast. Noted that pre-training stage would take a lot of times, so it needs to consider another way to speedup. [2014-06-22]

  6. Image Reconstruction indicated that this implementation is performed OK on average. So that if we only employ first layer's learning result, it should still give us state-of-art performance with a plain autoencoder. [2014-06-22]

  7. This is the major updates since the last one. In this version, Python support by SWIG is added. And now the Python library can be accessed under /build/Python. During this duration, I tested current implementation on complete CIFAR-10 dataset several times. By using different settings and pre-processings. With plain DeSTIN hierarchy, it is not capable to effectively classify images in tiny images. Basically, I conducted my experiments on most of set objectives. Surprisingly, when we apply DeSTIN to Robot behaviour learning studies, it shows some positive results (later will extend this work to sequence behaviour learning). Currently, it can differentiate up to 9 scenes around the environment. With some proper training, it can also be extend to some larger number of scenes. This seems revealing the potentials of limited scene recognition. [2014-08-10]

##Contacts##

Hu Yuhuang

Advanced Robotics Lab

Department of Artificial Intelligence

Faculty of Computer Science & IT

University of Malaya

duguyue100@gmail.com

You can’t perform that action at this time.