Introduction to Deep Learning

• Machine Learning is Different
  • h(x)
  • questions machine learning can answer
  • process
• Building these things
  • features to vectors
  • mathematical representation
    • model
    • cost/reward function
  • optimizing the things
    • gradients
• Recognizing digits
  • Graph
    • IN: Constants, Placeholders, Variables
    • OUT: Session params
  • Run
• linear model
  • model + cost function
  • tensorflow example
• multi-layer linear model (MLP)
  • model + cost function
  • tensorflow example
• better architecture
  • convolutional neural network
  • tensorflow example

Samples

Loosely based on TensorFlow Examples