Instructor Notes

Ideas for episodes:

• Feature engineering
• Training on unbalanced datasets
  • stratified sampling.
• Time series prediction
• Explainable models (e.g. with LIME/SHAP)
• Neural Nets
  • Backpropegation (SGD, Adaptive Learning, Momentum)
  • Convolutional Neural Nets
  • Activation functions (Binary, ReLU, Sigmoid, Softmax)
• Semi-supervised learning
• Genetic Algorithms
• Word2Vec
• Model evaluation (precision, recall, F1
• Transfer learning
• Active Learning