Skip to content
This repository has been archived by the owner on Oct 29, 2018. It is now read-only.
/ python-destin Public archive

Python implementation of the DeSTIN deep learning perception system using the Theano library

Notifications You must be signed in to change notification settings

opencog/python-destin

Repository files navigation

PythonDeSTIN is a repo for the development of Python DeSTIN (PyDeSTIN). As a starting point:

The PyDeSTIN will have Four Classes: learning_algorithm, Node, Layer and Network The Classes will be placed in a nested fashion as follows: -> Network -> Layer -> Node -> learning_algorithm

Installation Instructions: -> Clone the repo by running git clone https://github.com/tedyhabtegebrial/PythonDeSTIN.git

-> Downoad Cifar dataset for training and testing Available at http://www.cs.toronto.edu/~kriz/cifar.html Download the python version

-> Edit loadData.py to modify the location of the Cifar directory/ where you placed the downloaded cifar dataset -> run testDestin.py _______________________________ Nessesary Libraries See http://deeplearning.net/software/theano/install_ubuntu.html#install-ubuntu for installing it on Ubuntu You need to have python>2.73, Numpy and Scipy libraries installed For the future versions theano will also be necessary so installing theano is optional at this time

Testing Inorder to run the testWithSVM.py script and evaluate the classification accuracy on the cifar data set install the scikit-learn machine learning toolkit. see installation instructions at http://scikit-learn.org/stable/install.html

Outlines for the Development of DeSTIN as a robust Spatio-Temporal Inference Engine Taking into consideration points listed @: http://wiki.opencog.org/w/New_DeSTIN_Redesign_Proposal We will have explicit branches for A to D.

A) pure DeSTIN Framework: flexible enough to support different learning algorithms (Done)

B) Implemeting Online-NonNegative Sparse auto_encoder in theano/or python (Done)

C) Implemeting Stable Incremental K Means Clustering in theano (In Progress)

D) a LeNet style CNN built using the general-purpose CNN layer ()

(The theory may require revision)
(How to make sense of the Complex and Simple cell like filters simulated in 	the CNN?)
(Pooling is also an issue.....)

D) hybrid DeSTIN-CNN without feedback

D) hybrid DeSTIN-CNN with feedback

Reading List For DeSTIN:

About

Python implementation of the DeSTIN deep learning perception system using the Theano library

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 4

  •  
  •  
  •  
  •  

Languages