Location: EBW 240

Time: Wed & Fri, 11:00 - 12:15, fall semester, 2014

Office hours: Wed and Fri 4 - 5, EBW 109

Acknowledgement: The course development is supported by an National Science Foundation CAREER award. Some images and figures used in the lectures are provided by images.google.com and other sources.

Syllabus

Lectures

1. Introduction

2. Markov Chain Monte Carlo Methods

Reading assignment 1: read the first 22 pages of Introduction to MCMC methods for machine learning, write a half-page summary, submit it to mumachinelearning@gmail.com. Due on Sept. 5.

3. Iterative Improvement Algorithms (Hill Climbing, Simulated Annealing, and Genetic Algorithm)

4. Dynamic Programming

5. Linear and Integer Programming

6. Quadratic Programming with Application to Kernel Methods

Reading assignment 2: read the first 13 pages of Introduction to Kernel methods, write a half-page summary, submit it to mumachinelearning@gmail.com. Due on 14.

7. Contrastive Divergence Optimization with Applications in Deep Learning Networks

Reading assignment 3: read the paper "G.E. Hinton, S. Osindero, Y. Teh. (2006). A fast learning algorithm for deep belief nets. Neural Computation 18(7):1527-1554." , write a half page summary, and submit it to mumachinelearning@gmail.com (due on Dec. 12).

Projects

1. Search DNA sequence motif using MCMC

(Discussion of Plan (Sept. 10, Wed), Presentation of Plan (Sept. 12, Fri), and report on 9/19 (Friday)); Motif data set (Reference: Brown et al., MEME-LaB: motif analysis in clusters. Bioinformatics, 2013). One sample file (You can test your program on a few sequence files to search motifs with different lengths (6 - 15 nucleotides)).