CS 685 - Machine Learning: Statistical and Computational Foundations

Extracting meaningful patterns from random samples of large data sets. Statistical analysis of the resulting problems. Common algorithm paradigms for such tasks. Central concepts: VC-dimension, Margins of classifier, Sparsity and description length. Performance guarantees: Generalization bounds, data dependent error bounds and computational complexity of learning algorithms. Common paradigms: Neural networks, Kernel methods and Support Vector machines, Applications to Data Mining.