Informations

Course beginning:Febr. 26, 2020
Duration 13 weeks.

Lesson timetable

Tuesday 10:00 - 14:00

Room 20, SPV

Wednesday 08:00 - 12:00

Room 20, SPV

Tools

• didaProg   (FTP download)
• Letture_es_Matlab

Links

• Research Index CiteSeer
• SPIB
• Journal of Machine Learning Research
• ICA Central
• TC Blind Signal Processing  IEEE CAS Society
• TC Machine Learning for Signal Processing IEEE SP Society

International labs link

• Blind Singal Processing Links

Motivation

The main objectives of the course “Machine Learning” ( ML) is to introduce students to state-of-the-art methods and basic programming tools for analysis of complex and big-data set. The ML course provides theoretical and practical basic-tools for the study and determination of adaptive and machine learning algorithms: linear, nonlinear, supervised, unsupervised for static (or memoryless) applications: regression, classification and clustering; and dynamic application: adaptive filtering, modeling and prediction of complex physical phenomena. In particular, are provided: the theoretical and practical basic-tools for linear adaptive filtering in different application; the formal statistical methods for performance analysis of ML algorithms and the basic tools for the algorithm development on vector and parallel architectures and large-scale distributed environment. A secondary objective of the ML course  is to teach students how to develop, and evaluate, simple ML or adaptive filtering algorithms in various application domains such as: multimedia and multimodal communications, biological, biomedical, sound, telecommunications, remote sensing, social networks, internet, big-data, industry 4.0,  etc. Therefore, every student will lead a project that is based on machine learning state-of-the-art research.

Objectives of the Course

The student acquires the basic theory and is able to design, implement and evaluate the performance of, most common machine learning algorithms, also on parallel machines with different grain on several application contexts.

Main topics

Introduction to Adaptive Computation and Machine Learning

 

Bayesian Approach to Adaptive Computation

 

Basic Supervised and Unsupervised Algorithms for Pattern Recognition, Regression and Clustering

 

Solution of Underdetermined Systems of Linear Equations with Minimal L1- L2- Norm

 

Multilayer Neural Networks

 

Recurrent Neural Networks

 

Kernel Method and Regularized Networks

  

Dynamic Stochastic Neural Networks and Probabilistic Graphical Models

 

Deep Neural Networks Architecture, Learning and Applications

 

References

Textbooks

• Aurelio Uncini, “Introduction to Adaptive Algorithms and Machine Learning”, Ed. 2020,  available at A4Z copy center, via della Polveriera, 13A Roma (near SVP), 

• Aurelio Uncini, “Mathematical Elements for Machine Learning”, Ed.  2020 (free pdf available only for the students).

• Aurelio Uncini, "Fundamentals of Adaptive Signal Processing" - Springer, ISBN 978-3-319-02806-4, Feb. 2015  
• S. Scardapane, D. Comminniello, M. Scarpiniti, A. Uncini, “Designing Lerge Machine Learning Simulations Using the Lynx Toolbox".

Other recommended reading

• Kevin P. Murphy, “Machine Learning: A Probabilistic Perspective”,  Adaptive Computation and Machine Learning series, MIT Press book
• Sergios Theodoridis, “Machine Learning: A Bayesian and Optimization Perspective”, 1st Edition - Elsevier 2015.
• Ia Goodfellow and Yoshua Bengio and Aaron Courville, "Deep Learning", MIT Press book, Ed 2018.
• Dimitri P. Bertsekas and John N. Tsitsiklis, “Parallel and Distributed Computation: Numerical Methods,” ISBN 1-886529-01-9
• Rumelhart, D.E., Hinton, G.E., & McClelland, J.L. (1986). “A General Framework for Parallel Distributed Processing”, In Rumelhart, D.E., & McClelland, J.L. and the PDP Research Group (1986) Eds. Parallel Distributed Processing: Explorations in the Microstructure of Cognition. Volume 1: Foundations. MIT Press: Cambridge, MA.