Machine Learning

CSCI-B555 - Spring 2015 - section 26329


Class Meets

When: Tuesdays and Thursdays 1:00pm-2:15pm

Where: Lindley Hall 008



Predrag Radivojac

Office: LH301F



Office Hours (instructor)

Tuesdays 3pm-4pm

Thursdays 4pm-5pm or by appointment in LH301F


Associate Instructor(s)

Chenyou Fan


Office hours: Mon and Wed, 2:30-4:00, Informatics West 202


Course Objective

The course objective is to study the theory and practice of constructing algorithms that learn (functions) and make optimal decisions from data and experience. Machine learning is a field with goals overlapping with other disciplines, in particular, statistics, algorithms, engineering, or optimization theory. It also has wide applications to a number of scientific areas such as finance, life sciences, social sciences, or medicine.



Graduate student standing or permission of the instructor.




Pattern Recognition and Machine Learning - by C. M. Bishop, Springer 2006.



Machine Learning - by Tom M. Mitchell, McGraw-Hill, 1997

The Elements of Statistical Learning - by T. Hastie, R. Tibshirani, and J. Friedman, 2009


Supplementary material will be provided in class.


Topics: about 75% of the following topics depending on the year

▪  mathematical foundations of machine learning

▫  random variables and probabilities

▫  probability distributions

▫  high-dimensional spaces

▪  overview of machine learning

▫  supervised, semi-supervised, unsupervised learning

▫  inductive and transductive frameworks

▪  basics of parameter estimation

▫  point estimates

▫  Bayesian formulation

▫  Expectation-maximization algorithm

▪  classification algorithms: linear and non-linear algorithms

▫  perceptrons

▫  logistic regression

▫  naive Bayes

▫  decision trees

▫  neural networks

▫  support vector machines

▪  regression algorithms

▫  least squares linear regression

▫  neural networks

▫  relevance vector machines

▪  kernel methods (taught within classification and regression)

▫  dual representations

▫  RBF networks

▪  graphical models

▫  Bayesian networks

▫  Markov random fields

▫  inference

▪  ensemble methods

▫  bagging

▫  boosting

▫  random forests

▪  practical aspects in machine learning

▫  data preprocessing

▫  overfitting

▫  accuracy estimation

▫  parameter and model selection

▪  special topics (if time permits)

▫  introduction to PAC learning

▫  sample selection bias

▫  learning from graph data

▫  learning from sequential data



Midterm exam: 25%

Final exam: 25%

Homework assignments: 30%

Class (mini) project: 20%


Late Policy and Academic Honesty

All assignments and exams are individual, except when collaboration is explicitly allowed. All the sources used for problem solution must be acknowledged, e.g. web sites, books, research papers, personal communication with people, etc. Academic honesty is taken seriously; for detailed information see Indiana University Code of Student Rights, Responsibilities, and Conduct.

Last updated: 01/13/2015