Multimedia Information Systems
Part of this Programme
Information Technology Management
Level of Qualification|Semesters|ECTS
Bachelor | Semestral | 6
Year | Type of course unit | Language
3 |Mandatory |Português
Total of Working Hours | Duration of Contact (hours)
168 | 52,5
Recommended complementary curricular units
Prerequisites and co-requisites
- In-depth knowledge of the concepts associated with the information extraction life cycle in multimedia systems. - What concrete criteria differentiate data information - How a representation captures unexplained characteristics in the data - How do most multimedia data end up being represented as matrices - Capture and essential representation of images, sound, video, bio-data, and others. - Basic statistics for pre-processing of data: inference of missing values, standardization, normalization, distribution analysis, modality, central tendency, variance, symmetry, discrepant points (out of series), correlation, etc. - Analysis of data using NNMF (Non-Negative Matrix Factorization) - Analysis of data using PCA (Principal Component Analysis)
The central objective of this course is to provide students with algorithmic / mathematical techniques for extracting characteristics with high discriminatory / informative value from data generated in multiple mediums: image, sound, video, bio-data, etc. In this sense, there is a central focus on the life cycle of multi-media systems: (1) data acquisition; (2) selection of characteristics; (3) representation and (4) 'machine learning'. In this basic level chair the attention is more directed to points (1) and (2) but without losing sight of the fundamental elements of (3) and (4). In terms of skill set, students are prepared for (1) statistical pre-processing of matrix data for later (2) feature selection using matrix factorization with non-negative matrix factorization and main component analysis techniques. In addition, this chair requires self-didactic learning of the basic aspects of programming in Python and PHP.
Teaching methodologies and assessment
Classes in theory and practice. Classes always begin with a review of previous class concepts and questions to challenge students in cognitive terms. Usually there is an open problem, in which we work as the chair progresses. The evaluation is done with two tests in the theoretical class, and two practical works, one in PhP + SQL and another in PCA, and NNMF statistics.
- Eidenberger, H. (2012). Handbook of multimedia information retrieval. BoD¿Books on Demand. - Tukey, J. W. (1977). Exploratory data analysis.
- (practical) https://cran.r-project.org/web/packages/NMF/NMF.pdf
- (practical) http://factominer.free.fr/classical-methods/principal-components-analysis.html
- (article analysis) Ludeña-Choez, J., Quispe-Soncco, R., & Gallardo-Antolín, A. (2017). Bird sound spectrogram decomposition through Non-Negative Matrix Factorization for the acoustic classification of bird species. PloS one, 12(6), e0179403.