Machine Learning
Code
11157
Academic unit
Faculdade de Ciências e Tecnologia
Department
Departamento de Informática
Credits
6.0
Teacher in charge
Pedro Manuel Corrêa Calvente Barahona, Susana Maria dos Santos Nascimento Martins de Almeida
Weekly hours
4
Total hours
58
Teaching language
Português
Objectives
Knowledge
- Understand the paradigms and challenges of Machine Learning, distinguishing Supervised, Unsupervised and Reinforcement learning.
- Learn the fundamental methods and their applications in data oriented knowledge discovery. Understand data features, the selection of models and their complexity.
- Understand the advantages and disadvantages of the different methods.
Applications
- Implement and adapt Machine Learning algorithms;
- Model real data experimentally.
- Interpret and evaluate experimental results.
- Validate and compare different Machine Learning algorithms.
Soft Skills
- Evaluate the suitability of each method to concrete applications and data sets.
- Critical evaluation of the results.
- Autonomy and self-reliance in the application and furthering studies in Machine Learning.
Subject matter
1. Introduction
1.1 Supervised Learning, Unsupervised Learning and Reinforcement Learning.
1.2 Classification, regression and clustering.
1.3. Machine Learning in the context of Data Mining: applications
2. Data
2.1 Types of Data
2.2 Measures of similarity/dissimilarity and measures of data scatter.
2.3 Introduction to data pre-processing and visualization
2.4. dimensionality reduction
3. Supervised Learning
3.1 Regression
3.2 Decision Trees
3.3 Artificial Neural Networks
3.4 Support Vector Machines
3.5 Graphical models
3.6 K-nearest neighbour classifier
3.7 Methods for classifier evaluation and comparison
3.8 Ensembles
4. Unsupervised Learning
4.1 Partitional clustering
4.2 Hierarchical clustering
4.3 Self organizing maps
4.4 Probabilistic clustering
4.5 Fuzzy clustering
4.6.Clustering evaluation
Bibliography
- T. Mitchell. Machine Learning, McGraw-Hill, 1997.
- C. M. Bishop. Pattern Recognition and Machine Learning, Springer, 2006.
- E. Alpaydin. Introduction to Machine Learning, Second Edition, MIT Press, 2010.
- T. Hastie, R. Tibshirani, J. Friedman. Elements of Statistical Learning, Second Edition, Springer, 2009.
Evaluation method
The evaluation of this curricular unit is made by two components: theoretical/problems (T) and project (P). Each component contributes with 50% to the final grade.
Both components are evaluated in an integer scale from 0 to 20.
To pass, the student must have:
(i) a grade of at least 9,5 points in the theoretical/problems component; and
(ii) a grade of at least 9.5 points in the project component;
The final grade is defined as the weighted average of the two components of evaluation, that is (0.5×T + 0.5×P), in an integer scale from 0 to 20 points.
Theoretical/problems component
This component is evaluated by 2 written tests. The grade of this component is the average of the 2 written tests.
Alternatively, this component can be evaluated by a written exam.
Project component
This component is evaluated by 2 programming mini-projects accompanied by a written report.
The mini-projects will be mostly developed in the practical classes, each having a strict deadline to be delivered plus the corresponding report. Please, consult the dates below.
The mini-projects are done in group of students and have a unique grade (in a scale of 0 to 20 points).
Though the mini-projects are done in groups, the evaluation of this component is individual.
Grading of the different evaluation components are presented as a numerical value rounded to the first decimal place. The final grade is rounded to the closest integer value.
Students who made the project component in the last academic year (2013/2014) are allowed not to do it this year. In this case, the evaluation of this component is the one that they obtained previously.