Statistical Methods in AI - Monsoon 2018-19

The course serves as an introduction to statistical methods used in machine learning and AI. The topics covered includes pattern representation, classification, and clustering. A good working knowledge of Linear Algebra and Probability and Statistics is expected from the students as a prerequisite.

Topics

  • Introduction, Feature Representation
  • Random Variables, Probability Densities, Multivariate Densities
  • Linear Discriminant Functions
  • Perceptron Learning
  • Minimum Squared Error Procedures
  • Bayesian Decision Theory
  • Naive Bayes Classifier
  • Maximum Likelihood Estimation (MLE)
  • Machine Learning Fundamentals
  • Logistic Regression; Feature Selection
  • Principal Component Analysis and Eigen Faces
  • Linear Discriminant Analysis and Fischer Faces
  • Nearest Neighbour Classifier
  • Max-Margin Classification (SVM), SVM variants, Kernalization
  • Neural Networks, Backpropagation, Training Methods
  • Data Clustering, Kmeans (EM) and variants, Hierarchical Clustering
  • Decision Trees, Random Forests
  • Graphical Models, Bayesian Belief Networks
  • Combining Classifiers, Boosting.

Reference Books

  • Pattern Classification by Duda, Hart & Stork
  • Neural Networks by Simon Haykin
  • Machine Learning - A Probabilistic Perspective by Kevin Murphy (free ebook available online)

Scribes