Advanced machine learning for computer vision

Code:

MVCOMP06

Acronym:

ACAVC

Keywords
Classification	Keyword
CNAEF	Informatics Sciences

Instance: 2022/2023 - 2S

Active?	Yes
Responsible unit:	Department of Electrical and Computer Engineering
Course/CS Responsible:	Master in Computer Vision

Cycles of Study/Courses

Acronym	No. of Students	Study Plan	Curricular Years	Credits UCN	Credits ECTS	Contact hours	Total Time
MVCOMP	4	Syllabus	1	-	6	42	162

Teaching language

English
Obs.: Lecionada por docentes das Universidades de Vigo e da Corunha

Objectives

This curricular unit aims at introducing the students to advanced computational learning techniques, being an extension of the computational learning curricular unit studied in the previous semester.

Learning outcomes and competences

Ability to develop machine learning systems depending on existing needs and apply the most appropriate technological tools.
Know, apply and evaluate advanced learning models.
Know deep learning techniques, with end-to-end training approaches, and minimization of the use of tagged data.
Solve applications using advanced auto-learning methods.
Acquire the learning skills that allow to continue studying in a way that will be largely self-directed or autonomous.
Ability to work in team, organization and planning.
Ability to analyze and synthesize knowledge.

Working method

À distância

Program

Deep Learning. Deep Models.
Deep Convolutional Neural Ne tworks. Regularization techniques. Optimization techniques. End to end training.
AutoEncoders. Generative Models. Interpretability. Sparsity of models.
Applications in computer vision.
Data efficient learning. Weakly supervised learning (semi-supervised learning; zero-shot learning; one-shot learning; open class classification).
Multitask learning. Transfe r Learning. Active learning.
Applications in computer vision. Models for sequential data. Hidden Markov Models.
Recurrent Neural Networks.
Applications in computer vision. Reinforcement learning.
Applications.

Mandatory literature

Ian Goodfellow, Yoshua Bengio, Aaron Courville; Deep Learning, MIT Press, 2017
Edition. Richard S. Sutton and A ndrew G. Barto.; Reinforcement Learning, An Introduction., MIT Press, 2017
Sergios Theodoridis; Machine Learning: A Bayesian and Optimization Perspective., Academic Press, 2015

Teaching methods and learning activities

Participatory lectures, seminars and conferences, learning ba sed on the resolution of practical cases and projects, autonomous work and independent study by students, group work and cooperative learning.
Subjects will be covered both in participatory lectures, where students will have the chance t o implement methods for themselves. During the lecture part, the course topics will be presented and discussed. The practical/lab periods will be used for solving exercises and for the development of the assignments.

Evaluation Type

Distributed evaluation with final exam

Assessment Components

Designation	Weight (%)
Participação presencial	30,00
Exame	35,00
Trabalho prático ou de projeto	35,00
Total:	100,00

Amount of time allocated to each course unit

Designation	Time (hours)
Estudo autónomo	40,00
Frequência das aulas	42,00
Apresentação/discussão de um trabalho científico	10,00
Trabalho escrito	20,00
Trabalho laboratorial	50,00
Total:	162,00

Eligibility for exams

To be defined.

Calculation formula of final grade

Students will be assigned weekly individual homework assignments during the whole duration of the course, involving exercises, readings and summarization of selected texts. They will account for 30% of the final grade.Practical work will consist of one project covering the course topics. This will account for 35% of the final grade. The final exam will account for 35% of the final grade.