Download Pattern Recognition Winter 2015/2016 Andrew Cohen acohen@coe

Pattern Recognition
Winter 2015/2016
Andrew Cohen
[email protected]
What is this course about?
This course will study state-of-the-art techniques for analyzing data. The goal is to extract
meaningful information from feature data. This includes statistical and information
theoretic concepts relating to machine learning, data mining and pattern recognition, with
applications using MATLAB.
This course is intended for computer science and engineering graduate students, but is open
to any student with a background in probability and calculus. One additional requirement
is some background in programming (preferably including courses on data structures and
algorithms) and the willingness and ability to learn MATLAB.
Course meeting time
Tuesday/Thursday 11:30-12:20, Bossone 605
Tuesday/Thursday 2:30-3:20, Bossone 605
Course website
All course information including lecture notes and homework (found as the last slide in the
weekly lecture notes) will be posted on the course website :
http://bioimage.coe.drexel.edu/courses/PatternRecognition
Important: prior to the second lecture you must verify that you can access this syllabus document
on the course website.
Textbook
The required text for this course is
Theodoridis, S. and K. Koutroumbas, Pattern recognition. 4th ed. 2009, San Diego, CA:
Academic Press.
There is an additional MATLAB reference available for this text that you may find helpful.
Pattern Recognition Syllabus
Page 1 of 5
Note that you also need access to the MATLAB software, including the image processing and
statistics toolkits.
Instructor information
Andrew R. Cohen, Ph.D.
Office: Bossone Room 110
Lab: Bossone 514
e-mail: [email protected] (preferred mode of contact)
office phone: (251)571-4358
http://bioimage.coe.drexel.edu
office hours: by appointment, Tuesdays 3:30-4:30
TA information
Xaime Rivas Ray
Office: TBD
e-mail: <[email protected]> (preferred mode of contact)
office hours: Thursdays, 4-5, 3101 Market st, Room 232A
Prerequisite
An advance course on probability and statistics / probability for engineers is a required prerequisite. If
you are unsure if you meet this requirement, contact Prof. Cohen immediately.
Course Topics
(subject to change based on schedule / pace of lectures)

Bayes Decision Theory
o Discriminant Functions and Services
o the Normal Distribution
o Bayesian Classification
o Estimating Probability Density Functions
o Nearest Neighbor Rules
o Bayesian Networks

Linear Classifiers
o the Perceptron Algorithm
Pattern Recognition Syllabus
Page 2 of 5
o Least-Squares Methods

Nonlinear Classifiers
o Multilayer Perceptron's
o Back Propagation Algorithm
o Decision Trees
o Combinations of Classifiers
o Boosting

Feature Selection
o Data Preprocessing
o ROC Curves
o Class Separability Measures
o Feature Subset Selection
o Bayesian Information Criterion

Dimensionality Reduction
o Basis Vectors
o Singular Value Decomposition
o Independent Component Analysis
o Kernel PCA
o Wavelets

Additional Features And Template Matching
o Texture, Shape and Size Characterization
o Fractals
o Features For Audio
o Template Matching Using Dynamic Time Warping and Edit Distance

Context Dependent Classification
Pattern Recognition Syllabus
Page 3 of 5

Clustering
o Sequential Algorithms
o Hierarchical Algorithms
o Functional Optimization-Based Clustering
o Graph Clustering
o Learning Clustering
o Clustering High Dimensional Data


Subspace Clustering
Cluster Validity Measures
Grading
•
Assignments (roughly one per week) – 40%
•
Term Project – 50%
•
In class problem sets – 10%
•
Class Participation – 3%
•
Generous reward for effort
Term Projects
The lectures are designed to give you a broad understanding of the subject. The term project is
designed to give you an in-depth understanding of a selected topic that relates your research to a
topic from the course. The final project will include a written proposal. At the end of the
semester, each student will give a short oral presentation on their final project. Finally, due on
the date of the scheduled final exam will be a research paper describing your project. All of the
requirements for the term projects will be detailed later in the semester.
Pattern Recognition Syllabus
Page 4 of 5
Attendance
Attendence is mandatory. Please do not attend if you are ill. Excused absences must be reported
via e-mail. If you're not ill, you are expected to attend and participate in the lectures. Discussion
is a key aspect of learning, class participation is expected and will be tracked as part of your
grade.
Homework Submission
Homework should be neatly formatted, and include a write up (Summary, Conclusions, etc.) as
well as any code for programming problems. The writeup should be submitted as a pdf with a zip
file containing a code supplement. The PDF writeup should “stand alone”, with the code used
only if necessary.
Unless otherwise noted, homework should be submitted by email to the TA.
Homework is due no later than 5pm on the assigned due date. Homework that is late will have
10% deducted for every day it is late. IMPORTANT: all homework should be completed
individually. If you need help with the homework, contact Prof. Cohen.
Note that in your homework write-ups, neatness counts. Do a nice professional job on the
homework.
Academic conduct
University policy on academic conduct can be found here:
http://drexel.edu/studentaffairs/community_standards/studentHandbook/general_information/code_of_conduct/
IMPORTANT: PRIOR TO THE NEXT LECTURE BE SURE TO READ THIS POLICY
Pattern Recognition Syllabus
Page 5 of 5