Machine Learning

Course topics

Part 1. Intro, ML Overview

  • Class logistics
  • Project guidelines
  • Intro and history of ML
  • Overview of supervised learning
  • Overview of unsupervised learning
  • Overview of reinforcement learning

Part 2.  Linear Regression, Gradient Descent

  • Linear regression
  • Gradient descent
  • Normal equations

Part 3. Weighted Linear Regression, MLE, Logistic Regression

  • Weighted linear regression
  • MLE
  • Logistic regression

Part 4.  Newton, GLM, Softmax

  • Generalized linear models (GLM)
  • Linear regression and logistic regression from as GLMs
  • Newton’s method

Part 5. Generative Models, GDA, Naive Bayes

  • Generative models
  • Gaussian discriminant analysis
  • Naive Bayes

Part 6.  Regularization, Model Selection, Troubleshooting

  • Regularization
  • Model Selection
  • Troubleshooting

Part 7. SVM, Kernels

  • Margins
  • Max-margin classifier
  • Support vector machines
  • Kernels
  • Mercer’s theorem

Part 8.  Neural Networks

  • Highly non-linear class boundary
  • Neural network architecture
  • Activation function
  • Weights
  • Forward propagation
  • Backpropagation
  • Other useful activation functions

Part 9. Deep Learning

  • Learning features from data
  • Autoencoder
  • Feature composition
  • Popular deep learning models
  • Convolutional neural networks
  • Recurrent neural networks

Part 10.  Decision Trees, Ensemble, Bagging, Boosting

  • Decision trees
  • Ensembles
  • Bagging
  • Boosting
  • Random forest

Part 11.  Bias, Variance, Hypothesis Space, VC-Dimension

  • Bias-variance trade-off
  • Main theoretical principles
  • Finite hypotheses space
  • Infinite hypotheses space

Part 12. K-means, Mixture of Gaussians, EM

  • Unsupervised learning
  • Clustering
  • K-means
  • Gaussian Mixture Model
  • EM

Part 13.  PCA

  • Dimensionality reduction
  • PCA
  • Eigenfaces

Part 14. ICA

  • Cocktail party problem
  • ICA ambiguities
  • Densities and linear transformations
  • ICA algorithm

Part 15.  RL, MDP, Value, Policy

  • Reinforcement learning
  • Markov decision process
  • Value iteration
  • Policy iteration

Part 16. MDP Learning

  • MDP with discrete states
  • MDP with continuous states

Part 17.  Q-Learning, DQN

  • Q-learning for finite MDP
  • Q-learning with function approximation
  • Deep Q-learning

Part 18. Natural Language Processing

  • NLP applications
  • Classical techniques
  • Word embeddings
  • Language modeling
  • Deep Learning & NLP

Part 19.  Recommender Systems

  • Neighborhood-based recommender systems
  • Content-based recommender systems
  • Model-based recommender systems

Part 20.  Career in ML

  • ML in industry
  • ML in academia
  • ML in a startup
  • … not only ML

Prerequisites