CNN

##Here's some basic overviews:
http://cs231n.github.io/convolutional-networks/
http://deeplearning.net/tutorial/lenet.html

##Here's the TensorFlow MINST pages:
https://www.tensorflow.org/versions/r0.11/tutorials/mnist/beginners/index.html
https://www.tensorflow.org/versions/r0.9/tutorials/mnist/pros/index.html

---

Multilayer Convolutional Neural Network (Increases accuracy drastically from a single layer)

Branch

multi_cnn

OUTLINE

• Weight & Bias Variable Initialization
  • Function for Weight
  • Function for Bias
• Convolution & Pooling
  • Function for stride size when using convolutions
  • Function for max pooling
• First Convolutional Layer
  • Set Weight and Bias
  • Reshape the image into a 4d tensor
  • 2nd and 3rd dimension correspond to width and height
  • 4th dimension corresponds to color channels
  • Convolve x_image with the weight tensor, add the bias
  • Max_Pool the convolve
• Second Convolutional Layer
  • Now instead of 32 features for each 5x5 patch it will now have 64 features
• Densely Connected Layer
  • Add in a fully-connected layer to allow processing on entire image
  • Tensor from the pooling layer is reshaped
  • Dropout is then applied
• Readout Layer
  • Softmax layer is added
• Train & Evaluate the Model