| Code |
15660
|
| Year |
2
|
| Semester |
S1
|
| ECTS Credits |
6
|
| Workload |
PL(30H)/T(30H)
|
| Scientific area |
Informatics
|
|
Entry requirements |
Does not have.
|
|
Learning outcomes |
The course aims to introduce students to the fundamentals, methods and applications of Machine Learning (ML).
The techniques, design methodologies and algorithms that illustrate supervised, unsupervised and reinforced approaches to ML are discussed.
Different algorithms and computational techniques used in modern ML are presented and the respective experimental application to real data is exemplified.
At the end of this course students should be able to:
- Understand the paradigms and challenges of the area of ML.
- Understand the motivations, assumptions and limitations of the various computational techniques that are applied to solve a particular problem in ML.
- Explore the existing implementations of the most popular ML algorithms and know how to implement and adapt them properly.
- Identify directions for research in ML.
- Show autonomy in the adoption and adaptation of ML techniques.
|
|
Syllabus |
1. Introduction.
Preliminary Concepts and Definitions: Supervised Learning, Unsupervised Learning, and Reinforcement Learning.Areas of application. Examples of classification, regression, and grouping problems.
2. Data processing topics.
Distances and similarities. Measures of dispersion and visualization of data. Dimensionality reduction. Detection of anomalies.
3. Supervised Learning.
Regression. Classification. Instance-based classifiers. Graphic models. Neural Networks. Decision Trees. Support Vector Machines. Ensembles.
4. Unsupervised Learning.
Clustering Analysis. Partitioning methods. Probabilistic clustering methods. Fuzzy clustering methods. Hierarchical clustering methods. Self-organized maps.
5. Selected applications.
Control systems, recommendation systems and large-scale machine learning.
|
|
Main Bibliography |
David Barber, Bayesian Reasoning and Machine Learning, Cambridge University Press, 2016.
Ian Goodfellow, and Yoshua Bengio, Deep Learning, MIT Press, 2016.
Andrew Ng, Machine Learning Yearning, 2018.
E. Alpaydin (2010). Introduction to Machine Learning, Second Edition, MIT Press.
Scientific papers and extra materials provided by the teacher.
|
|
Teaching Methodologies and Assessment Criteria |
Theoretical classes followed by discussion of the presented algorithms, complemented with small theoretical and practical exercises. Practical classes with the use of computers and conducting group work. Students have the opportunity, with teacher guidance, to apply learned techniques to problems with real data within a small scale software development project. It is also planned to carry out an individual presentation on an area of recent research in ML.
The knowledge assessment component (1 test and a pair learning assignment) has a weight of 50% of the final mark, the assessment of practical procedures has a weight of 45%, the remaining 5% weights the student's ability to participate and discuss. Compulsory attendance is required in 80% of classes.
|
|
Language |
Portuguese. Tutorial support is available in English.
|