ECS 256: Probability Models for Computer Science

ECS 256
Probability Models for Computer Science
Effective Term
2019 Spring Quarter
Learning Activities
Lecture - 3.0 hours
Project (Term Project)
Extensive Problem Solving
Probabilistic and statistical models useful in computer/data science. Applications to networks, bioinformatics, database management, machine learning, software engineering and image processing. Not open for credit to students who have completed ECS 256A.
A calculus-based course in probability, such as ECS132, STA 131A, or EEC 161; programming skills and familiarity with matrix algebra.
Credit Limitation
Not open for credit to students who have completed ECS 256A.
Enrollment Restrictions
Pass One and Pass Two open to graduate students in Computer Science only.

Summary of Course Content


I.  Brief review of prerequisite material in probability (0.5 weeks).

II.  Discrete event simulation, using either the 'simmer' (R) or SimPy
(Python) languages (2 weeks).

III.  Discrete and continuous-time Markov chains (1.5 weeks).

IV.  Queuing models (1.5 weeks).

V.  Application of Markov models to classification techniques:  
Hidden Markov models (1 week).

VI.  Application of Markov models to Bayesian methods:
Markov Chain Monte Carlo (1 week).

VII.  Specific computer science applications, case studies (2.5 weeks).


Illustrative Reading

Instructor's materials. Online textbook can be found:


Potential Course Overlap
This course does not have a significant overlap with any other course.  It covers some topics similar to some in course 271, but with a more statistical view.  There is also some topic similarity to MAT 135B, but with a much more applied emphasis. The major theme of this course, running throughout Topics II-VI above, is Markov models, applied in computer science contexts. There are no courses at all in the Department of Statistics on Markov modeling,  MAT 135B does cover Topic III, but in a much more theoretical manner and not with computer science applications.  Topic V has some overlap with course 271, but with a more statistical view. Topic V also has some overlap with STA 208, but again uniquely motivated by Markov models. 

Note:  This course originally was titled Probabilistic Modeling of Computer Systems.  At some point, it was re-titled Performance Evaluation, but has always been taught in the original form, involving Markov models and applications to queuing analysis.  The current change would revert to (close to) the original title, with a modernized topic list, notably adding Hidden Markov Models, Markov Chain Monte Carlo, and discrete event simulation software.

Final Exam
Yes Final Exam

Justification for No Final Exam
Grading: Letter; homework (50%), project (25%); final exam (25%) The project will include the design and analysis of a computer and/or communication system using the analytical and simulation methodologies developed in this course.

Course Category