Artificial Intelligence Topics in Medicine: Machine Learning

Recommended Prerequisites

NA

Teaching Methods

Teaching methods are essentially expository and dialectical, but also resorting to the resolution of practical problems. The contextualization of the problems and concepts are initially presented to students, allowing a discussion of possible applications in medicine. Subsequently, a practical problem to be solved by the students is introduced. Thus, the ability to apply and interpret the results obtained is developed, which is enhanced by the use of appropriate computer tools.

Learning Outcomes

To understand the fundamentals of unsupervised learning.

To plan unsupervised learning applications to medicine.

To recognize the methods, advantages and disadvantages of reinforcement learning.

To identify the main aspects of natural language processing.

Work Placement(s)

No

Syllabus

1. Unsupervised learning

    i) Applications

    ii) Types of conglomerates

    iii) Algorithms: dbscan, mean-shift, k-means, hierarchical

2. Reinforcement learning

    i) Essential elements of the algorithm.

    ii) Formulation of a reinforcement learning problem

    iii) Q-learning and SARSA.

    iv) Applications

3. Natural language processing

    i) Techniques used in NLP

    ii) Applications.

Head Lecturer(s)

Francisco José Santiago Fernandes Amado Caramelo

Assessment Methods

Assessment
Laboratory work or Field work: 100.0%

Bibliography

- Christopher M. Bishop, Pattern Recognition and Machine Learning, Springer

- Mário Rodrigues, António Teixeira, Advanced Applications of Natural Language Processing for Performing Information Extraction, Springer

- Richard S. Sutton and Andrew G. Barto, Reinforcement Learning: An Introduction, A Bradford Book, The MIT Press.