#MNIST handwritting recognition using TensorFlow

This repository is for the implementation of handwritting recogition with MNIST dataset using TensorFlow.

#Get Started

To run the programs please install TensorFlow with Python 2.7, using the instructions given here

For running basic softmax regression run the following command:

# Using default configuration for logs directory, batch size and number of iterations:
$ python softmaxReg.py

# For custom configuration use the following command:
$ python softmaxReg.py --max_it <no. of iterations> --batch <batch size> --log_dir <path to logs directory>

# For tensorboard visualization run following command and  visit http://locahost:6006:
$ tensorboard --logdir <path to logs directory>

#Data Flow Graphs

##Basic Softmax Implementation

###Overview Graph

###Expanded Graph

#Results

##Basic Softmax Implementation

Following results were obtained with batch size = 100 and number of iterations = 1000

###Accuracy

###Cross Entropy

###Weights Visualization

####0:

####1:

####2:

####3:

####4:

####5:

####6:

####7:

####8:

####9: