Neural networks basics (slides)

  • The classification problem- again
  • NN history
  • Perceptron
    • Hyperplanes
    • Activation
  • Dense layer
  • Multi-layer perceptron (MLP)
  • Optimization
    • Softmax + cross entropy + loss
    • Gradient descent
  • Basic data preprocessing
    • Data normalization
    • Train, validation and test splits
  • Fully connected net