The course addresses advanced topics in machine learning applied to the fields of the acquisition and processing of images and computer vision. The corse builds upon the machine learning fundamentals studied in the other courses to explore cutting-edge deep learning topics. This includes transformer-based foundational models for multimodal processing, and neural networks for computational imaging problems like super-resolution. It also delves into generative models for image synthesis, including the latest diffusion-based approaches, and continual learning techniques for their personalization.

The course will provide students with a solid knowledge of state-of-the-art deep learning models in computer vision and their real-world adoption. The course will also provide students with practical abilities to design, use and customize such techniques. For this purpose, each course topic is coupled with computer labs where students are expected to implement and test algorithms and learn the effect of various parameters. In particular, the course will allow the student to achieve the following results. 1. Knowledge of adaptive non-local models for vision and image processing (graph neural networks, Transformers) 2. Knowledge of modern multi-modal foundational models for visual understanding (CLIP, ...) 3. Knowledge of computational imaging principles and neural networks for solution of inverse problems (denoising, super-resolution, ...) 4. Knowledge of image generative models (GANs, denoising diffusion models) 5. Knowledge of continual learning techniques for personalization of generative models (finetuning, Adapters, LoRAs, ...)

Students are expected to have basic knowledge of machine learning principles, simple neural network designs (fully-connected layers, CNNs, normalization layers, ...) and their training (loss functions, backpropagation algorithm). Basic knowledge of probability and statistics (discrete and continuous random variables, probability distributions, ...) is also expected. Basic concepts will be reviewed at the beginning of the course. Python programming is required for computer labs.

- Review of neural networks (1h) - Graph signal processing and graph neural networks (6h) - Applications to sensor networks, IoT, image analysis - Transformers and transformer-based architectures for image, video, time series processing (9h) - Neural networks for computational imaging (9h) - Inverse problems in image processing (denoising, super-resolution, ...) - Generative models for image and video synthesis (11h) - Unconditional vs. text-conditional models - GANs, Diffusion models - Continual learning and personalization (Finetuning, Adapters, LoRA, ...)

The course will consist in 36 hours of lectures and 24 hours of labs. Lab activities will cover the course topics and each will last around 3-6 hours. Students will work in teams for the lab assignments. Lab reports are expected and will be part of the final grade.

All topics will be covered by the course slides. Extra reading material will consist in research papers highlighted in course content.

Slides;

Modalità di esame: Prova scritta (in aula); Elaborato scritto prodotto in gruppo;

Exam: Written test; Group essay;

... The exam aims at verifying the knowledge and understanding of the topics discussed during the course, and the ability of the students to critically discuss such topics. The final exam will consist of two parts. The first part will be a written exam. It lasts 1.5 hours and consists in discussing up to 3 topics, each topic discussion having limited size (one page of text). The written exam contributes to the final score for up to 24 points, and the discussion of each topic usually contributes equally to the score. During the written exam, students are not allowed to use any books, lecture notes or any material. They must avoid having any active cell phone, tablet or other electronic means. The second part will be the evaluation of the lab reports. This score will be up to 9 points and it will be awarded based on the technical merit of the work done, as well as the clarity and completeness of the report presentation (including the understanding of the concepts learned during the course as applied to each lab activity).

Gli studenti e le studentesse con disabilità o con Disturbi Specifici di Apprendimento (DSA), oltre alla segnalazione tramite procedura informatizzata, sono invitati a comunicare anche direttamente al/la docente titolare dell'insegnamento, con un preavviso non inferiore ad una settimana dall'avvio della sessione d'esame, gli strumenti compensativi concordati con l'Unità Special Needs, al fine di permettere al/la docente la declinazione più idonea in riferimento alla specifica tipologia di esame.