Title: Applications of Modern
Statistical Learning Theory in Networked Embedded Systems
Course catalog number: ELEC 697 (Fall 2006)
Instructor: Dr. Farinaz Koushanfar,
Office Hours: TR after the regular class and by email appointments
Meeting time: 2:30 PM - 3:50PM TR
Meeting place: Room 2014 DH
Prerequisites: Prior knowledge of undergraduate-level probability and statistics is a plus, but the course is self-contained
Announcements | Syllabus | Course Outline | Lectures and Handouts | Projects | Reading List
Announcements:
·
(10/2/06) There will be
no class next Tuesday, Nov 7
·
The official class number
will be changed to ELEC 536!
·
The project proposal
sessions (presentations) will be on Thursday 9/19/06 and Tuesday 9/24/06
·
(9/12/06) The
introduction to R class will be held next Tuesday after the regular class
hours, from 4:00-5:00PM.
·
(8/23/06) The
organizational meeting will be held on Tuesday,
August 29, 2:30-3:30PM, Room 2014, Duncan Hall
·
(8/23/06) Course flyer for ELEC 697
· Overview
o Practical statistical learning methods and tools
o Modeling and optimizing emerging embedded systems
o Research areas: networked embedded systems and a lot of sensor networks
o
Emphasizing the methods and their foundation and
usage rather than the theoretical aspects
· Goals
o Solid understanding of the state-of-the-art learning methods
o Hands-on experience with statistical modeling packages
o Applications of statistical modeling in Ad-hoc networks, embedded systems, SN
o A universal tool for your own research, depending on your interest, the class project may be on a diverse set of topics, including, but not limited to: sensor networks, intrusion detection in Internet, and VLSI statistical manufacturing variability
· Textbook
o The elements of statistical learning: data mining, inference, and prediction, by T. Hastie; R. Tibshirani; J. Friedman; New York : Springer, 2001.
· Recommended further reading
o Pattern Classification (2nd ed.), by R. Duda; P. Hart; D. Stork; Wiley Interscience, 2001.
o Modern Applied Statistics with S-PLUS, Third Edition, W. Venables; B. Ripley; Springer, 1999.
o Papers from the literature
· Course webpage
o www.ece.rice.edu/~fk1/classes/ELEC697.htm
· Grading
o Weekly assignments (20%)
o Mid-semester oral presentation (15%)
o Paper presentation and discussion (15%)
o Class project report (30%)
o Class project presentation (20%)
· Project
o Groups of 1 or 2
o Need dataset to analyze and model
o Either propose or select from my projects/datasets
· Modeling/data analysis software
o S programming language (Splus/R)
o You can download R from CRAN at: http://cran.us.r-project.org/
o Documentation is also available at CRAN
o Many more resources available on the web
Week 1: Orientation and overview of supervised learning and its applications in embedded networks
Week 2: Intro to R, Linear regression, model selection, validation
Week 3: Applications of regression in embedded networks (HW 0)
Week 4: Linear classification: LDA, logistic, separating hyperplanes
Week 5: Applications of classifications in embedded networks (HW 1)
Week 6: Available datasets, possible project proposals, and project selection
Week 7: Model assessment and selection
Week 8: Applications of models selection and validation in embedded networked systems (HW 2)
Week 9: Kernel methods
Week 10: Applications of kernel methods in embedded networked systems (HW 3)
Week 11: Mid-term project proposal and presentations
Week 12: Model inference and averaging: boosting, ML, EM
Week 13: Applications of model inference in embedded networked systems (HW4)
Week 14: Progress report -- presenting the related work to your project and your goals
Week 15: Summary
Week 16: Final project presentation and reports (Report)
Lecture 1: Introduction and organization, applications in networked embedded systems and sensor networks (Slides)
Lecture 2: Overview of data analysis methodologies (Slides) (papers)
Lecture 3: Overview of supervised learning, least square and nearest neighbor methods (Slides)
Lecture 4 and 5: Linear regression, confidence intervals, multiple regressions, model selection and shrinkage methods (Slides)
Homework 0: (due on Tuesday, Sept 26 in class); Note that, the problem set is only to evaluate your background and understanding, it will not be graded but I would really appreciate your feedback; (HW0-R)(HW0-problems)(dataset)
Lectures 8 and 9: Linear Classification, LDA, logistic regression, separating hyperplanes (Slides)
Marco Duartes Presentation: Classification and tracking of vehicles on a sensor networks testbed (Slides)
Homework 1: (due on Thursday, Oct 26 in class); (HW1-R)(HW1-problems)(dataset)(learn set)(test set)
Lecture 10: Kernel methods (Slides)
Lecture 11 and 12: Model assessment and selection (Slides)
Lecture 13 and 14: Model inference: ML, Bayesian, EM, MCMC, bagging (Slides)(Tutorial on EM)
Lecture 15 and 16: Additive models, trees, and related methods (Slides)
Lecture 17 and 18: Boosting and additive trees (Slides)
Traffic characterization in ad-hoc networks
Statistical characterization of low-power embedded sensing network
Statistical testing for detecting the presence of recombination
Metering integrated circuits via statistical manufacturing process variations
Secure compressive sensing