Catalogo dei Corsi di studio

General Objectives:

The objectives of this course are to present a wide spectrum of Machine
Learning methods and algorithms, discuss their properties, convergence
criteria and applicability. The course will also present examples of
successful application of Machine Learning algorithms in different
application scenarios.
The main outcome of the course is the capability of the students of
solving learning problems, by a proper formulation of the problem, a
proper choice of the algorithm suitable to solve the problem and the
execution of experimental analysis to evaluate the results obtained.

Specific Objectives:

Knowledge and understanding:
Providing a wide overview of the main machine learning methods and
algorithms
for the classification, regression, unsupervised learning and
reinforcement learning problems. All the problems are formally defined
and theoretical basis as well as technical and implementation details
are provided in order to understand the proposed solutions.

Applying knowledge and understanding:
Solving specific machine learning problems starting from training data,
through a proper application of the studied methods and algorithms. The
development of two homeworks (small projects to be developed at home)
allows the students to apply the acquired knowledge.


Making judgements:
Ability of evaluating performance of a machine learning system using
proper metrics and evaluation methodologies.

Communication skills:
Ability of writing a technical report describing the results of the
homeworks, thus showing abilities in communicating results obtained from
the application of the acquired knowledge in solving a specific problem.
Being exposed to examples of communication of results obtained in
practical cases given by experts within seminars offered during the course.

Learning skills:
Self-study of specific application domains, problems and solutions
during the homeworks, with possible application of teamwork for the
solution of specific problems.

Programme

Classification

Basic concepts and evaluation
Decision Trees
Bayes Learning
Linear Models
Support Vector Machines
Kernels
Multiple classifiers

Regression

Linear and logistic regression
Instance based (K-NN)
Perceptron
Neural networks
Deep neural networks (CNN)

Unsupervised learning

Clustering (k-Means)
Latent variables (EM)

Reinforcement learning

MDP
Q-learning
Deep Reinforcement Learning
HMM
POMDP

Adopted texts

Machine Learning, Tom Mitchell.
Pattern Recognition and Machine Learning, Chris Bishop.
Machine Learning: a Probabilistic Perspective, Kevin Murphy.
Introduction to Machine Learning, Ethem Alpaydin.
Deep Learning, Ian Goodfellow, Yoshua Bengio, Aaron Courville.
Reinforcement Learning, Sutton and Barto.

Prerequisites

Basic knowledfge of programming languages, algorithms, data structures, probability and statistics.

Exam modes

The exam consists in a written test.