Teaching plan for the course unit

 

 

Close imatge de maquetació

 

Print

 

General information

 

Course unit name: Probabilistic Graphical Models

Course unit code: 569428

Academic year: 2021-2022

Coordinator: Jerónimo Hernández González

Department: Department of Mathematics and Computer Science

Credits: 4,5

Single program: S

More information enllaç

 

 

Estimated learning time

Total number of hours 112.5

 

Face-to-face and/or online activities

45

(Due to the Covid-19 restrictions, we expect to have 50%-75% of in-person activities)

Supervised project

37.5

Independent learning

30

 

 

Learning objectives

 

Referring to knowledge

- To know what probabilistic graphical models (PGMs) are and what queries we can ask them.    

- To know when (and how) queries can be answered exactly in polynomial time (exact inference), and what to do when they can not (approximate inference).

- To know the basic techniques to learn probabilisitic graphical models from data, even when data is partially missing.

 

Referring to abilities, skills

- To be able to apply PGM algorithms to problems of your interest.

- To be able to translate PGMs and related algorithms into code.

 

 

Teaching blocks

 

1. Representation

1.1. Introduction to PGMs

1.2. Bayesian networks

1.3. Markov networks

1.4. Plate models

2. Inference

2.1. Exact inference

2.2. Approximate inference

3. Learning from data

3.1. Learning the model parameters

3.2. Building the graph structure

3.3. Learning when data is partially missing

4. Modern trends, applications and tools

 

 

Teaching methods and general organization

 

Teaching will follow a face-to-face (in-person), virtual (online) or mixed model according to the instructions of the competent authorities. In principle, we expect to follow the mixed teaching model for the 2021-22 academic year. 

* In case of in-person teaching:

Lectures dynamically combine master explanations and problem solving.  The weekly schedule of in-person activities is distributed in three hours. Some slots may be exclusively dedicated to programming throughout directed activities or notebooks.

The students will be required to present the application of PGMs (their own or other people’s) to problems of their interest, or a recently proposed PGM technique.

* In case of mixed teaching required by the health situation (this is the expected model):

If the health situation allows it and the necessary conditions are met, we expect to have between 50% and 70% of in-person activities. In general, when having an occupancy rate of 50%, students will attend in-person for a week and will follow class on streaming for the following week.

* In case on-line teaching is required by the health situation:

The time ranges of mixed teaching are maintained but all teaching will be carried out in an online format, prioritizing synchronous sessions. 

As far as possible, the gender perspective will be incorporated in the development of the subject.

 

 

Official assessment of learning outcomes

 

The subject is expected to be evaluated based on a final exam (40%), a presentation (30%) and in-class activities (30%).

Depending on the health situation, the evaluable activities can be in-person or virtual.

 

Examination-based assessment

The subject is expected to be evaluated based on a final exam (50%) and a project (50%).

Depending on the health situation, the evaluable activities can be in-person or virtual.

 

 

Reading and study resources

Consulteu la disponibilitat a CERCABIB

Book

Daphne Koller, Nir Friedman (2009). Probabilistic graphical models: principles and techniques. MIT Press, Cambridge (MA), USA.

https://cataleg.ub.edu/record=b1952061~S1*cat  Enllaç