CSE 627 Machine Learning (3 credits)

Catalog description:

Concepts and algorithms of machine learning including version-spaces, decision trees, instance-based learning, networks, evolutionary computation, Bayesian learning and reinforcement learning.

Prerequisite:

CSE 274 / CSE 606 or equivalent.

Course Objectives:

  • Describe and demonstrate the major paradigms of machine learning algorithms and knowledge structures.
  • Articulate the strengths and weaknesses, the appropriateness of various algorithms and the tradeoffs involved in machine learning implementations.
  • Implement software using one or more machine learning approaches.
  • Study and critique machine learning research literature.

Required topics (approximate weeks allocated):

  • Introduction to machine learning (0.5)
  • Concept learning (1.0)
    • Concept learning as search
    • Version spaces
  • Decision trees (2.0)
    • Representation
    • Decision tree learning algorithm
    • Inductive bias
  • Neural networks (2.0)
    • Perceptron learning rule
    • Multi-layered networks
    • Backpropagation learning algorithm
  • Bayesian Learning (1.5)
    • Bayes theorem
    • Bayes optimal classifier
    • Bayesian networks
  • Instance-based learning (1.5)
    • K-nearest neighbor classifier
    • Case-based reasoning
    • Radial basis functions
  • Evolutionary computation (1.5)
    • Genetic algorithms
    • Genetic programming
  • Reinforcement Learning (1.0)
    • Q Learning
    • Temporal Difference Learning
  • Applications and subtopics of machine learning -- examples are: (3.0)
    • Analytical Learning
    • Combining Inductive and Analytical Learning
    • Computational Learning Theory
    • Ensemble learning
    • Evaluating Hypotheses
    • Evolutionary programming and evolutionary strategies
    • Fuzzy Learning
    • Learning Sets of Rules
    • Machine learning applied to games
    • Unsupervised learning
  • Exams/Review (1)