Schedule
Monday, February 19
Morning (9:00 - 13:00 including 30-minute coffee break)
Opening ceremony
Dimensionality reduction for quantitative data – Part I (Maurizio Vichi)
Afternoon (14:00 - 18:00 including 30-minute coffee break)
Dimensionality reduction for quantitative data – Part II (Maurizio Vichi)
The two lectures provide an introduction to supervised and unsupervised learning. The tentative syllabus is:
- partitioning and hierarchical methods with a modelling approach;
- dimensionality reduction methods (Principal Component Analysis, Exploratory Factor Analysis, Confirmatory Factor Analysis and Structural Equation Modeling - SEM);
- sequential and simultaneous clustering and multidimensional reduction.
Tuesday, February 20
Morning (9:00 - 13:00 including 30-minute coffee break)
Dimensionality reduction for categorical data – Part I (Michael Greenacre)
The lecture provides an overview of Correspondence Analysis and related methods. The tentative syllabus is:
- correspondence analysis (CA) as the categorical equivalent of Principal Component Analysis;
- dimension reduction and clustering of cross-tabulations, frequency data, proportions and ratio-scale data in general;
- the logarithmic, Box-Cox and logratio transformations.
Afternoon (14:00 - 18:00 including 30-minute coffee break)
Dimensionality reduction for categorical data – Part II (Michael Greenacre)
The lecture focuses on Multiple Correspondence Analysis and related methods. The tentative syllabus is:
- multiple correspondence analysis (MCA) of multivariate categorical data;
- dimensionality and total variance of multivariate categorical data;
- subset MCA;
- clustering of categories in a multivariate categorical dataset;
- categorical Principal Component Analysis (catPCA).
Wednesday, February 21
Morning (9:00 - 13:00 including 30-minute coffee break)
Fuzzy unsupervised classification (Paolo Giordani)
The lecture focuses on fuzzy techniques for unsupervised classification. The tentative syllabus is:
- fuzzy logic;
- fuzzy extensions of the k-means algorithm;
- fuzzy extensions of the k-medoids (also labelled Partitioning Around Medoids, PAM) algorithm;
- fuzzy clustering of mixed data.
Afternoon (14:00 - 18:00 including 30-minute coffee break)
Model-based unsupervised classification (Roberto Rocci)
In the lecture the finite mixture model is introduced and discussed to implement a model-based approach to the unsupervised classification problem. The tentative syllabus is:
- definition and main properties;
- maximum likelihood estimation and the EM algorithm;
- finite mixture of Gaussians;
- outliers;
- choice of the number of components.
Thursday, February 22
Morning (9:00 - 13:00 including 30-minute coffee break)
Model-based classification (Roberto Rocci)
In the lecture the finite mixture model is evolved to cover cases where the analysis intention is not completely unsupervised. The tentative syllabus is:
- finite mixture of regression models;
- finite mixture of experts;
- semi-supervised classification;
- supervised classification.
Afternoon (14:00 - 18:00 including 30-minute coffee break)
Supervised Classification (Agostino di Ciaccio)
The lecture introduces nonlinear supervised classification techniques, widely used in machine learning. The tentative syllabus is:
- introduction and application fields of machine learning;
- evaluation methods in classification;
- ensemble methods based on classification trees;
- neural networks;
- supervised classification with neural networks.
Friday, February 23
Morning (9:00 - 13:00 including 30-minute coffee break)
Ongoing research (Lecturers present their current research interests)
Afternoon (14:00 - 18:00 including 30-minute coffee break)
Ongoing research (Participants, upon request, may present their current research interests)
Closing ceremony