Machine Learning Seminar Course,

CS6501 - 02

Department of Computer Science, University of Virginia

Fall 2013

Course Logistics

Using a collaborative learning style, this course aims to enhance the CS graduate students' understanding of statistical machine learning methods.  

Class Description: The course will take the form of a seminar instead of lectures by the instructor. We will go through the textbook, one chapter per class except that a heavy chapter may be split into two or three classes. Each class starts by summarizing questions from all the participants about the current chapter, followed by a presentation (lecture) on that chapter, and then classroom discussions about collected and new questions. Students will be grouped into teams of two or three persons; each team is assigned to two chapters or three).  Each team will analyze, deliver a lecture and lead the classroom discussions. All the students are required to read every chapter before it is discussed in a class, and sent their questions through Collab in advance.

Class Time: Tuesdays & Thursdays 11:00am-12:15pm , starting on Tuesday August 27th, 2013

Class Location: Rice Hall 120

TA: Tung Dao ( thd7wk@virginia.edu) will hold the office hours from 4-5pm on every Tuesday.

Instructors:

• Yanjun (Jane) Qi, Rice Hall 503

Course Website :

• Collab page

• Corrections or comments to yq2h@virginia.edu


Tentative schedule:

Date	Content
TU Aug 27	Overview of class
TR Aug 29	Overview of logistics
TU Sept 3	Ch 1-2. Introduction; Supervised Learning
TR Sept 5	Ch 3. Linear Methods for Regression (part 1) 3.1-3.4.4
TU Sep 10	Ch 3. Linear Methods for Regression (part 2) 3.5-3.9 (no 3.7)
TR Sep 12	Ch 4. Linear Methods for Classification (4.1-4.4)
TU Sep 17	Ch 5. Basis Expansion and Regularization (5.1 to 5.5)
TR Sep 19	Ch 6. Kernel Smoothing Methods
TU Sep 24	Ch 7. Model Assessment and Selection (7.1 – 7.10)
TR Sep 26	Ch 8. Model Inference and Averaging (8.1-8.7)
TU Oct 1	Ch 9. Additive Models, Trees, and Related Methods (9.1-9.4)
TR Oct 3	Special topics / Review/ Sample Projects
TU Oct 8	Ch 10. Boosting and Additive Trees (part 1:  10.1 – 10.9)
TR Oct 10	Ch 10. Boosting and Additive Trees (part 2:  10.10 – 10.14)
TR Oct 17	Ch 11. Neural Networks
TU Oct 22	Ch 12. Support Vector Machines & Flexible Discriminants (part 1: 12.1-12.3.8 + Ch 4.5 )
TR Oct 24	READING DAY
TU Oct 29	Ch 13. Prototype Methods and Nearest-Neighbors
TR Oct 31	Ch 14. Unsupervised Learning (part 1:  14.1 and 14.3 )  + Ch 12.5 - note: no 14.2 in this class !
TU Nov 5	Ch 14. Unsupervised Learning (part 2:  14.5, 14.6, 14.7 )
TR Nov 7	Ch 14. Unsupervised Learning (part 3:  14.2, 14.4, 14.8-14.10 )
TU Nov 12	Ch 15. Random Forest + Ch 8.7,8.8,8.9
TR Nov 14	Ch 16. Ensemble Learning
TU Nov 19	Ch 17. Undirected Graphical Model
TR Nov 21	Ch 18. High-Dimensional Problem (Part 1: 18.1-18.4.2)
TU Nov 26	Ch 18. High-Dimensional Problem (Part 2: 18.5-18.7)
TU Dec 3  (Rice 120)	Project Presentation by each student ( 7 minutes / student)
WED Dec 4  (Rice 120)	Project Presentation by each student ( 7 minutes / student)
THUR Dec 5 (Rice 120)  (11am - 12:30pm)	Project Presentation by each student ( 7 minutes / student)
Mon Dec 9

Textbook:

• Textbook (required): Elements of Statistical Learning, Hastie, Tibshirani and Friedman.

• www.stanford.edu/~hastie/local.ftp/Springer/OLD//ESLII_print4.pdf

Question from each student before each class:

• For each class' content, please submit at least 3 questions.

• TA will grade the questions as assignments.

Grading:

• No exams in this course.

• Sit-in: No.  This course is for registered students only.

• 2 pages write-up required for permitted absence of a class

•

• Final grades will be based on.

• –15% for participation in class;

• –15% for the quality of the questions and discussion in class;

• –40% for the quality of the seminar presentations;

• –30% for the quality of written project reports by each individual in the second half of the semester

Course Project Report:

• By the end of this semester, each student is required to submit a project report (through three incremental phases).

• This report (IEEE conference paper submission format, double column, 4 pages min limit, 10 pages max limit).

• Template could be obtained from: http://www.ieee.org/conferences_events/conferences/publishing/templates.html

• It aims to motivate students applying statistical machine learning on their research projects, or some research problem they are interested with.

• At the same time, on the last three classes, each student is required to make an in-class 7 minutes presentation about their course proposal.

• The project proposal is due around the mid of Oct (2 pages length requirement).

• The mid-phase project report is due in the mid of Nov.  

• The final draft of this proposal is due at the end of this semester.

• Depending on the student’s level of ML, your report could be either a proposal or a full project report (extra credits will be added if full project report).