Click here for our new site

Convolutional Neural Nets and Automatic Differentiation

Chris Van Horne

This will be a 10,000 foot view of artificial neural networks (ANNs), the convolution, convolutional neural networks (CNNs), and automatic differentiation. ANNs will be shown graphically: what they look like structurally, what they can recognize, what they can create after being trained, backpropogation and gradients. Finally, we will cover a neat technique called automatic differentiation for calculating exact gradients used during backpropogation learning as opposed to the traditional difference quotient or symbolic manipulation methods.

Fork me on GitHub