The project is to implement Supervised Discrete Hashing algorithm on MNIST dataset and calculate precision and recall on the same.

System Requirements This was tested on sytem with Intel Core I5, 8GB Ram, 64 bit Ubuntu 16.04 LTS Python 2.7 was installed on system with required libraries

Download link MNIST Dataset http://yann.lecun.com/exdb/mnist/

Run the following code - to convert mnist to pkl file python mnist_load.py

Run the following code - to create phi matrix for test and training data(you can edit the no of anchor points here) python phi.py

Run the following code - to run SDH and save B,P,W matrix (you can edit the codeLength here) python SDH.py

Run the following code - to collect statistics precision, recall and accuracy python -W ignore result.py

Conditions Folder Dataset should contain (Unzip Dataset.zip) t10k-images-idx3-ubyte t10k-labels-idx1-ubyte train-images-idx3-ubyte train-labels-idx1-ubyte Please ignore the warning SettingWithCopyWarning: A value is trying to be set on a copy of a slice from a DataFrame.

Run information of SDH.py and result.py is given in the python code as comments Dependency list is in file 'Dependecy List' Citations are given in file 'Citations'

Single File to run entire code for a code length is created It is run.py python -W ignore run.py This displays graph results of results we have stored(it displays current result in console)

This software is available under the following licenses: MIT

Link for complete code with dataset https://www.dropbox.com/s/npo4qqp90oyvtmi/FSL.zip?dl=0