The course will provide tools for the analysis of big data (audio, video, text) generated by modern information and communication systems and related services.
Skills stemming from computer science, statistics and optimization will be introduced to provide the means for understanding, designing and implementing methods capable of managing complex amounts of data, and transforming them into useful and semantically relevant information.
Topics to be discussed will include advanced principles of information theory (sparse coding, compressive sensing, random matrix), principles of statistical inference, methodologies for clustering the observed data, predictive analytics, and principles of constrained optimization based on elements of game theory.
Skills stemming from computer science, statistics and optimization will be introduced to provide the means for understanding, designing and implementing methods capable of managing complex amounts of data, and transforming them into useful and semantically relevant information.
Topics to be discussed will include advanced principles of information theory (sparse coding, compressive sensing, random matrix), principles of statistical inference, methodologies for clustering the observed data, predictive analytics, and principles of constrained optimization based on elements of game theory.
teacher profile teaching materials
Principles of statistical inference are introduced in the first part of the course.
The theory of the main machine learnign techniques, including regression,linear classification, and dimensionality reduction, is then illustrated.
Deep learning techniques are then detailed, discussing convolutional netural networks (CNNs), recurrent neural networks (RNNs), and combinations of both. also advanced concepts such as siamese networks, object detection, generative models and adversarial networks are treated.
Practical exercises using Matlab and Python will be performed to show the application of the considered techniques to real-world scenarios.
Programme
The course will provide tools for the analysis of big data (audio, video, text) generated by modern telecommunication systems and related services.Principles of statistical inference are introduced in the first part of the course.
The theory of the main machine learnign techniques, including regression,linear classification, and dimensionality reduction, is then illustrated.
Deep learning techniques are then detailed, discussing convolutional netural networks (CNNs), recurrent neural networks (RNNs), and combinations of both. also advanced concepts such as siamese networks, object detection, generative models and adversarial networks are treated.
Practical exercises using Matlab and Python will be performed to show the application of the considered techniques to real-world scenarios.
Core Documentation
Course slides.Reference Bibliography
S. Nolan and T. Heinzen, “Statistics for the Behavioral Sciences” G. James, D. Witten, T. Hastie, R. Tibshirani, “An Introduction to Statistical Learning” I. Goodfellow, Y. Bengio, A. Courville, “Deep Learning” S. Theodoridis and K. Koutroumbas, “Pattern Recognition” S. Theodoridis and K. Koutroumbas, “Introduction to Pattern Recognition - A Matlab Approach” K. P. Murphy, “Machine Learning - A Probabilistic Perspective” T. A. Runkler, “Data Analytics - Models and Algorithms for Intelligent Data Analysis”Type of delivery of the course
Lectures, seminars, exercisesType of evaluation
The verification of the acquired knowledge takes place through an oral exam, in which the level of students' effective understanding of the exposed concepts, and their ability to apply them in real contexts, is checked. If required, a written exam can be done.