Obiettivi formativi Learning goals
Statistical Machine Learning provides a rigorous introduction to the statistical and probabilistic principles that underpin modern artificial intelligence, including predictive and generative AI, and data-driven decision making. The course presents machine learning as a statistical discipline, where learning algorithms are viewed as methods for inference from data under uncertainty.
The course combines theoretical foundations with methodological and computational aspects. Students will develop an understanding of probabilistic modelling, statistical inference, optimization and generalization theory, together with the practical skills required to implement, evaluate and compare modern learning algorithms. Representative methodologies include kernel methods, deep neural networks, probabilistic generative models such as denoising diffusion probabilistic models (DDPMs), and modern uncertainty quantification techniques including conformal prediction. Throughout the course, particular emphasis is placed on principled model assessment, uncertainty quantification, and the interpretation of predictive models.
Theoretical results are complemented by programming assignments and a substantial project based on real datasets, allowing students to connect statistical principles with computational implementation using modern open-source scientific computing environments.
Knowledge and understanding
Upon successful completion of the course, students will understand the statistical principles underlying modern machine learning and their role in predictive and generative artificial intelligence. They will be able to interpret learning algorithms within a unified probabilistic framework, understanding how statistical modelling, inference, optimization and regularization jointly determine predictive performance and generalization.
Students will understand the assumptions, mathematical foundations and computational characteristics of the principal supervised and unsupervised learning methodologies, including kernel methods, (deep) neural networks and probabilistic generative models. They will also understand the theoretical principles governing model selection, validation, complexity control/regularization, and uncertainty quantification, together with the statistical tools used to assess and compare learning algorithms.
Finally, students will acquire familiarity with modern scientific computing environments for implementing, analysing and experimentally evaluating statistical machine learning methods.
Applying knowledge and understanding
Upon successful completion of the course, students will be able to formulate complex data analysis problems within an appropriate statistical learning framework and identify suitable modelling strategies based on the characteristics of the data and the objectives of the analysis.
They will acquire the practical skills required to implement, train and compare modern machine learning methods, integrating statistical modelling with computational techniques to build complete data analysis pipelines. Students will learn to assess predictive performance using principled statistical methodologies, quantify predictive uncertainty, and interpret the results in light of the assumptions and limitations of the adopted models.
Programming assignments and a final project will provide hands-on experience with real-world datasets, enabling students to bridge theoretical concepts with practical implementation and to develop reproducible machine learning solutions.
Making judgements
Upon successful completion of the course, students will have developed the ability to critically evaluate statistical learning methodologies from both theoretical and empirical perspectives. They will be able to assess whether the assumptions underlying a learning model are appropriate for a given application, compare competing modelling approaches, and interpret empirical results in light of statistical evidence.
Particular emphasis is placed on understanding the role of uncertainty in statistical learning, enabling students to distinguish between predictive performance, statistical significance, and confidence in model outputs. Students will develop the ability to critically assess the reliability, robustness and limitations of learning algorithms when applied to complex real-world data.
Communication skills
Students will develop the ability to communicate statistical and machine learning methodologies effectively through written reports and oral presentations. They will learn to present methodological choices, experimental results, and quantitative evidence in a clear, rigorous and reproducible manner, adapting the level of technical detail to both specialist and non-specialist audiences.
The final project will also foster the ability to discuss modelling decisions, interpret experimental findings, and provide constructive scientific feedback on the work of peers.
Learning Skill
The course aims to develop the independent learning skills required for continued study and research in statistical machine learning and artificial intelligence. By combining theoretical analysis with computational practice, students will acquire the ability to understand new methodologies from the scientific literature, critically evaluate their statistical foundations, and adapt them to novel application domains.
Particular emphasis is placed on developing a probabilistic mindset for reasoning under uncertainty, enabling students to understand emerging machine learning methodologies, integrate advances from the rapidly evolving AI landscape, and pursue lifelong learning with methodological rigour and intellectual independence.
|
Obiettivi formativi General objectives
The primary objective is to study algorithms and data structures that efficiently solve problems whose solution, using trivial approaches, would require very high resources. The algorithms studied will be implemented in Java language.
Specific objectives
Knowledge and ability to understand
Basic data structures and their use in solving sorting, searching, and graph problems will be shown. It will be shown how the data structures described are made available in Java language. The student will be able to determine the computational complexity of algorithms, and associate them with the appropriate complexity class.
Ability to apply knowledge and understanding
At the end of the course the students will be able to determine algorithms to efficiently solve complex problems, in particular problems on graphs, and choose the most suitable data structures to obtain an efficient implementation of the algorithm. They will also be able to use the Java classes that implement the data structures studied.
Autonomy of judgment
Students will be able to distinguish the computational complexity of problems and algorithms, and to identify the computationally more costly steps in solving a problem.
Communication skills
Students will be able to describe, in appropriate terms, the characteristics of the main data structures, and identify the primitives needed to efficiently implement an algorithm.
Learning ability
Students who pass the exam will be able to take advanced courses, of an applicative nature, that require the use of sophisticated algorithms. They will also be able to appreciate software engineering and computational complexity theory teachings.
|
Obiettivi formativi Learning goals.
The different techniques existing for Big Data management will be illustrated, with a particular emphasis on NoSQL databases.
The course will also deal with the problem of collecting Big Data from various sources such as from the web or from the online social networks.
This will require also the introduction of the different formats that are commonly used to encode unstructured, semi-structured and structured data and of the different techniques that can be used to automate their processing.
Successively, pre-processing techniques, including denoising and imputation of missing data, will be considered.
Then, the course will treat dimensionality reduction techniques, based on feature extraction and feature selection.
Finally, some statistical learning models, supervised and unsupervised, for the analysis of Big Data, will be presented.
Real-world problems will be addressed during the course using suitable software.
Knowledge and understanding.
The student will learn as to apply some statistical learning techniques for dimensionality reduction, based on feature extraction and feature selection.
Moreover, he will know and understand some powerful statistical learning models, supervised and unsupervised, to analyse Big Data.
Applying knowledge and understanding.
The student will be able to manage Big Data collected from various sources.
He will learn as to apply dimensionality reduction techniques, based on feature extraction and feature selection.
Moreover, he will be able to choose and apply some powerful statistical learning models to analyse Big Data.
Making judgements.
Students will develop critical skills through the application of a wide range of machine learning and statistical models.
They also will develop the critical sense through the comparison between alternative solutions to the same problem obtained using different learning logics.
They will learn to critically interpret the results obtained by applying the procedures to real data sets.
Communication skills.
Students, through the study and execution of practical exercises, acquire the technical-scientific language of the discipline, which must be used appropriately in both the intermediate and final written tests and in the oral tests.
Communication skills are also developed through group activities.
Learning skills.
Students who pass the exam will have learned an analytical approach that allows them to tackle Big Data analysis with statistical models and machine learning methods.
|
