Machine Learning II

Code:

CC3043

Acronym:

CC3043

Level:

300

Keywords
Classification	Keyword
OFICIAL	Computer Science

Instance: 2023/2024 - 1S

Active?	Yes
Responsible unit:	Department of Computer Science
Course/CS Responsible:	Bachelor in Artificial Intelligence and Data Science

Cycles of Study/Courses

Acronym	No. of Students	Study Plan	Curricular Years	Credits UCN	Credits ECTS	Contact hours	Total Time
L:BIOINF	0	Official Study Plan	3	-	6	48	162
L:IACD	61	study plan from 2021/22	3	-	6	48	162

Teaching language

Portuguese

Objectives

This UC consists of an introduction to some of the algorithmic foundations of deep and reinforcement learning.

It is intended that students have a first contact with such concepts and with concrete methods of implementing such algorithms.

They should be able to carefully select suitable algorithms and details of model architectures and learning techniques for each of the tasks presented.

They should know how to estimate the performance of the applied methods and use this information for iterative model design.

Learning outcomes and competences

- Understanding the fundamentals of deep learning and reinforcement learning algorithms and methodologies.
- Ability to justify the choice of a deep or reinforcement learning solution to a given problem
- Ability to design solutions to new problems based on deep or reinforcement learning methods
- Ability to evaluate the performance of the proposed solution and optimize the model

Working method

Presencial

Pre-requirements (prior knowledge) and co-requirements (common knowledge)

- Initial knowledge of artificial intelligence and data science
- Initial knowledge of computational learning (for example, attendance at the UC of Aprendizagem Computacional I).
- Programming knowledge preferably Python.
- Knowledge of basic matrix algebra and calculus in R and R^n

Program

- Introduction to neural networks and deep learning
- Training techniques
- Convolutional neural networks
- Recurring neural networks
- Generative models
- Reinforcement learning
- Elements of explainability in deep learning

Mandatory literature

Ian Goodfellow and Yoshua Bengio and Aaron Courville; Deep Learning, MIT Press, 2016

Complementary Bibliography

Daniel A. Roberts; The principles of deep learning theory. ISBN: 978-1-316-51933-2

Teaching methods and learning activities

In theoretical classes, the expository method will be used, presenting the themes of the program.

The practical classes will consist of the resolution of exercises to apply the concepts introduced in the theoretical classes and the development of a practical work.

Software

Python

Evaluation Type

Distributed evaluation with final exam

Assessment Components

designation	Weight (%)
Exame	40,00
Trabalho prático ou de projeto	40,00
Teste	20,00
Total:	100,00

Amount of time allocated to each course unit

designation	Time (hours)
Estudo autónomo	64,00
Frequência das aulas	48,00
Elaboração de projeto	50,00
Total:	162,00

Eligibility for exams

- Grade above 5 in practical work.

Calculation formula of final grade

F = 0.2*T + 0.4*E + 0.4*P

F: final grade
T: test
E: exam
P: practical work

In order to pass, students must have a grade higher than 7 points in the exam.

Special assessment (TE, DA, ...)

Students under special circumstances should describe their situation to the responsible.

Classification improvement

The exam grade and the test grade can be improved in the appeal season. The practical work grade cannot be improved after submission.