The class aims at providing the basic theories and the most significant methods related to the analysis of data of whatever nature, that is theory and methods related to pattern recognition and machine learning.
This discipline is at the base and complete many other disciplines, recently of larger diffusion, like image processing, the analysis of huge quantities of data, artificial intelligence, databases, and many others.
In tha class, special emphasis will be devoted to the probabilistic and statistical techniques, in particular to the learning of systems for classification and recognition.
Many applications are involved by this discipline.
To quote some, image analysis and computer vision, data mining, bioinformatics, biomedical image and biological data analysis and interpretation (e.g., genomics, proteomics, etc.), biometry, video surveillance, robotics, speech recognition, and many others.
* Introduction: what it is, what is useful for, systems, applications
* Recognition and classification
* Bayes theory
* Parameters' estimation
* Non parametric methods
* Linear classifiers, non linear classifiers, discriminant functions
* Feature estraction and selection, PCA and Fisher transform
* Expectation-Maximization e mixture of Gaussians
* Generative and discriminative methods
* Kernel methods e Support Vector Machines
* Artificial Neural Networks
* Hidden Markov Models
* Unsupervised classification (clustering)
In total, there are 32 hours of Theory lectures and 24 hours of laboratory.
An oral interview with 2 questions, aimed at verifying the understanding of theoretical concepts, and a project aimed at understanding the mastering of the mathematical and computer tools.
The oral test can be substituted with a written test with short questions similar to the oral one.
