Teaching code

4S009069

Teacher

Francesco Setti

Coordinator

Francesco Setti

Credits

6

Language

English

Scientific Disciplinary Sector (SSD)

ING-INF/05 - INFORMATION PROCESSING SYSTEMS

Period

Semester 1 dal Oct 2, 2023 al Jan 26, 2024.

Courses Single

Authorized

Lessons timetable Moodle Seminars 0

Learning objectives

The course aims to provide the basic tools for machine learning, together with specific techniques to deal with large amounts of data, such as deep learning. Theory and techniques will be specifically addressed to data science issues with particular emphasis on data analysis. At the end of the course the student has to show to have acquired the following skills:
- knowledge of the main types of data (e.g. binaries, texts, sounds, etc.)
- understanding and capability to use the basic elements of descriptive statistics, elementary probability, linear algebra with elements of optimization and regularization
- knowledge of basic machine learning techniques (e.g. support vector machines, random forest, etc.)
- knowledge of basic deep learning techniques (e.g. convolutional neural network, long-short memory machines, etc.)
- knowledge of the basics of Natural Language Processing for, for example, sentiment analysis
- knowledge of the basic issues in the context of measurement and Regression measures, e.g., RMSE (Root Mean Square Error), MAE, Rsquared and adjusted Rsquared)
- knowledge of the basic tools in supervised training, e.g., confusion matrix, accuracy, precision, recall, F1, Curve precision-recall, ROC, average precision, CMC NLP: Bleu, Spice

Prerequisites and basic notions

The student should have basic skills in math, linear algebra, probability and statistics.

Program

- Introduction to Machine Learning: basics, terminology, performance metrics, inductive bias
- Bayesian decision theory
- Parametric learning
- Support Vector Machines
- Artificial Neural Networks
- Convolutional Neural Networks
- Recurrent Neural Networks
- Auto-Encoders
- Unsupervised Machine Learning

Bibliography

Didactic methods

Lectures, exercises, laboratory sessions on the PC

Learning assessment procedures

The exam involves the discussion of a project proposing a solution to an industrial problem.
The student will present his/her work in about 15 minutes (with or without the use of support material such as slides, written report, demo, etc.), followed by a Q&A session.

Evaluation criteria

For the generation of the mark it will be taken into account:
- performance of the developed system (with different metrics depending on the problem);
- theoretical motivation behind the student's design choices;
- ability to clearly and concisely present the key points of the project;
- ability to support a discussion on possible alternative solutions and potential causes of failure of the solution developed.
The student must also demonstrate mastery of all the topics in the program (even those not addressed during the project).

Criteria for the composition of the final grade

The mark will be based on the discussion of an individual project focusing on the topics of the course.

Exam language

Inglese/English