• Basic information
• Learning outcomes
• Contents
• Assessment
• Workload

Learning outcomes

Upon successful completion of this course, students will be able to:

1. Use the fundamental deterministic inference methods and learning algorithms used in signal and parameter estimation.

   Contribution to Program Outcomes

   1. Define and manipulate advanced concepts of Electronics Engineering
   2. Formulate and solve advanced engineering problems
   3. Acquire scientific knowledge
   4. Find out new methods to improve his/her knowledge
   5. Effectively express his/her research ideas and findings both orally and in writing

   Method of assessment

   1. Written exam
   2. Homework assignment
   3. Term paper
2. Use the fundamental Monte Carlo inference methods and learning algorithms used in signal and parameter estimation.

   Contribution to Program Outcomes

   1. Define and manipulate advanced concepts of Electronics Engineering
   2. Formulate and solve advanced engineering problems
   3. Acquire scientific knowledge
   4. Find out new methods to improve his/her knowledge

   Method of assessment

   1. Written exam
   2. Homework assignment
   3. Term paper
3. Use the fundamental inference methods and learning algorithms used in linear and non-linear dynamical models

   Contribution to Program Outcomes

   1. Define and manipulate advanced concepts of Electronics Engineering
   2. Formulate and solve advanced engineering problems
   3. Acquire scientific knowledge
   4. Find out new methods to improve his/her knowledge

   Method of assessment

   1. Written exam
   2. Homework assignment
   3. Term paper

Contents
Week 1:	Random variables
Week 2:	Pairs of random variables
Week 3:	Random vectors
Week 4:	Gaussian random vectors, maximum likelihood (ML) estimation
Week 5:	Maximum-a-posteriori (MAP) and minimum mean square error (MMSE) estimations
Week 6:	Computational methods for estimation: Steepest descent, conjugate gradient and Newton's method
Week 7:	Monte Carlo methods for estimation: Rejection and importance sampling
Week 8:	Computational methods for multi-unkown models: Iterated conditional mode (ICM), coordinate descent
Week 9:	Midterm exam
Week 10:	LASSO, expectation-maximization (EM), Gibbs sampling
Week 11:	Generative and discriminative models for classification: Bayes classifier, logistic regression and softmax
Week 12:	Clustering: Expectation-maximization and k-means for finite mixture models
Week 13:	Random processes, Gauss-Markov processes, sequential Bayesian filtering
Week 14:	Kalman filter and particle filtering
Week 15*:	General review
Week 16*:	Final exam
Textbooks and materials:	C. M. Bishop, "Pattern Recognition and Machine Learning", Springer 2006.
Recommended readings:	G. Andrew, J.B. Carlin, H.S. Stern, D.B. Rubin, "Bayesian Data Analysis", Chapman & Hall/CRC, 2014.
	* Between 15th and 16th weeks is there a free week for students to prepare for final exam.

Assessment
Method of assessment	Week number	Weight (%)
Mid-terms:	9	30
Other in-term studies:		0
Project:	15	15
Homework:	3,5,7,10	15
Quiz:		0
Final exam:	16	40
	Total weight:	(%)

Workload
Activity	Duration (Hours per week)	Total number of weeks	Total hours in term
Courses (Face-to-face teaching):	3	14
Own studies outside class:	4	14
Practice, Recitation:	0	0
Homework:	5	8
Term project:	10	2
Term project presentation:	0	0
Quiz:	0	0
Own study for mid-term exam:	12	1
Mid-term:	3	1
Personal studies for final exam:	12	1
Final exam:	3	1
		Total workload:
		Total ECTS credits:	*
	* ECTS credit is calculated by dividing total workload by 25. (1 ECTS = 25 work hours)