Educational objectives Educational goals
The course provides students with fundamental knowledge of the operation of modern computing systems, from the system software that manages the resources of a single computer to the protocols that enable communication between hosts across the Internet. The course is organized into two integrated parts. The first part is devoted to operating systems and addresses the organization and services of the operating system, process and thread management, CPU scheduling, synchronization and concurrency, memory and virtual memory management, file systems, and the management of I/O and storage devices. The second part is devoted to computer networks and the Internet architecture, with particular emphasis on the TCP/IP protocol stack, the main protocols and mechanisms that enable host-to-host communication, packet forwarding, reliability and congestion control, name resolution, and medium access in local and wireless networks. Modern transport protocols, including QUIC, will also be presented, highlighting their role in the evolution of the Internet architecture. The course also includes practical activities aimed at understanding the behavior of the operating system and at configuring, observing, and analyzing networks and protocols.
Knowledge and understanding
By the end of the course, students will have acquired solid knowledge of the principles underlying operating systems and computer networks. Regarding operating systems, they will understand the role of the operating system as a resource manager and as an interface between hardware and applications, the life cycle of processes and threads, the main CPU scheduling policies, synchronization mechanisms and the classic problems of concurrency (mutual exclusion, semaphores, monitors, deadlock), as well as techniques for physical and virtual memory management, paging, and the organization of file systems and I/O. Regarding networks, students will understand the role of the different layers of the TCP/IP stack, from application-layer protocols to transport, network, and medium access mechanisms. They will understand the operation of fundamental protocols such as IP, TCP, UDP, and DNS, as well as the main routing mechanisms and issues related to reliability, congestion control, fragmentation, addressing, and channel sharing. In addition, they will acquire basic knowledge of wireless networks and the CSMA/CA protocol, and will understand the motivations and main characteristics of modern transport protocols such as QUIC.
Applying knowledge and understanding
Students will be able to analyze the behavior of an operating system, evaluating the effect of different process, memory, and resource management policies on overall system performance. They will be able to reason about concurrency scenarios, identifying and preventing race conditions, deadlocks, and consistency problems, and to compute fundamental quantities such as waiting times, completion times, and CPU utilization. Regarding networks, they will be able to analyze the operation of a TCP/IP network, identifying the role of the different protocols involved in end-to-end communication, interpret communication scenarios, and compute fundamental quantities such as delay, throughput, channel utilization, and transmission window sizes, evaluating the behavior of transport and network protocols under congestion, losses, or bandwidth constraints. They will also be able to configure simple network scenarios, observe packet exchanges, and interpret the results obtained through analysis tools and practical laboratory activities.
Making judgements
The course aims to develop students' ability to critically evaluate the design choices adopted both at the operating system level and at the network architecture level. Students will be able to discuss the advantages and limitations of the main solutions adopted in the management of a computer's resources, such as scheduling policies, synchronization mechanisms, and memory management strategies, and to compare network architectural solutions considering aspects such as scalability, reliability, efficiency, delay, security, mobility, and interoperability. They will also be able to compare traditional and modern protocols, such as TCP and QUIC, understanding the design motivations behind their evolution and evaluating the impact of different choices on the overall performance of the system and the network.
Communication skills
Students will acquire the ability to clearly and rigorously describe the operation of an operating system and of the main Internet protocols, as well as the mechanisms that regulate resource management and communication between devices. They will be able to use appropriate technical language to explain concepts such as process, thread, scheduling, synchronization, deadlock, virtual memory, encapsulation, addressing, routing, flow control, congestion control, name resolution, and medium access. The course also promotes the ability to present and discuss results derived from exercises, protocol analysis, and practical activities.
Learning skills
By the end of the course, students will have developed the skills needed to independently explore advanced topics in the fields of operating systems, computer networks, and the evolution of the Internet. They will be able to read technical documentation, understand system and protocol specifications, interpret experimental results, and update their knowledge with respect to the development of new systems, protocols, architectures, and technologies.
|
Educational objectives Educational Goals
The course, entitled AI Lab: Computer Vision & NLP and hereinafter referred to simply as AI Lab, aims to provide students with advanced knowledge and practical skills for the design, development, and evaluation of intelligent Artificial Perception systems based on the most recent advances in Artificial Intelligence. Although the course title explicitly refers to Computer Vision and Natural Language Processing (NLP), it should be interpreted from a broader perspective as an advanced Artificial Intelligence laboratory focused on the understanding of images, videos, language, and heterogeneous signals, where Computer Vision, signal processing, Machine Learning, Deep Learning, and state-of-the-art Artificial Intelligence models converge.
The course considers Computer Vision as its core discipline while extending its methodologies and principles to the analysis of heterogeneous information sources such as EEG, Wi-Fi sensing, physiological signals, and data acquired from heterogeneous sensing devices, demonstrating how many Artificial Intelligence paradigms can be transferred across different perceptual domains. In parallel, the course introduces elements of Natural Language Processing, highlighting how images, videos, language, and signals represent complementary forms of information through which intelligent systems can perceive, interpret, and understand the surrounding world. Particular attention is devoted to the main challenges of modern Artificial Perception, including image classification, object detection, semantic and instance segmentation, image registration, tracking, change detection, saliency detection, anomaly detection, action recognition, action anticipation, scene understanding, image captioning, visual reasoning, visual grounding, image retrieval, and other advanced Computer Vision tasks. The course also introduces methodologies for advanced signal analysis and fundamental concepts for the intelligent processing of non-visual data. To address these challenges, the course presents both classical Computer Vision and signal processing techniques as well as the most recent Artificial Intelligence paradigms. Students will be introduced to hand-crafted algorithms, Machine Learning techniques, Deep Learning architectures, and state-of-the-art AI models, including representative examples such as Convolutional Neural Networks (CNNs), Residual Networks (ResNet), U-Net and its variants, Siamese Networks, Teacher–Student architectures, Graph Neural Networks (GNNs), Vision Transformers (ViTs), Foundation Models, Vision-Language Models, and Large Language Models (LLMs). The objective is not merely to learn how these models work, but above all to develop the ability to critically understand their principles, advantages, limitations, and application scenarios.
A fundamental principle of the course is the development of a critical approach to the design of intelligent systems. Students will learn that there is no universally optimal algorithm and that the choice between classical approaches, hand-crafted algorithms, Machine Learning, Deep Learning, and Foundation Models depends on the nature of the problem, the available data, computational constraints, interpretability requirements, performance objectives, and the target application domain. Consequently, the course promotes a problem-driven approach, encouraging students to select the most appropriate methodology rather than relying indiscriminately on a specific technology. The course maintains a strong application-oriented perspective by demonstrating how the presented methodologies can be employed to develop intelligent systems for autonomous robotics, humanoid robots, unmanned aerial vehicles (UAVs), ground rovers, mobile robots, autonomous underwater vehicles (AUVs/UUVs), autonomous vehicles, satellite imagery and remote sensing, Human–Computer Interaction, biometrics, medical applications, environmental monitoring, industrial monitoring, smart cities, security, defense, ambient intelligence, and many other scientific and industrial scenarios in which the automatic perception, interpretation, and understanding of complex environments play a fundamental role. Finally, particular emphasis is placed on developing both scientific and engineering thinking. Rather than simply presenting established techniques, the course aims to train students capable of critically analyzing the international scientific literature, understanding the state of the art, comparing alternative methodologies, designing original solutions, and rigorously evaluating their experimental performance. The ultimate goal is to educate future engineers and researchers capable of developing innovative, scientifically sound, and internationally competitive Artificial Intelligence systems.
Knowledge and Understanding
Upon successful completion of the course, students will have acquired an in-depth understanding of the theoretical and methodological foundations of modern Artificial Perception, learning how images, videos, language, and heterogeneous signals can be processed through advanced Artificial Intelligence techniques to extract meaningful information for understanding humans, environments, and operational contexts. Students will understand the major challenges addressed by modern Computer Vision and advanced signal analysis, acquiring knowledge of image classification, object detection, segmentation, image registration, action recognition, motion analysis, change detection, saliency detection, anomaly detection, visual reasoning, image captioning, scene understanding, and other Artificial Perception problems. They will also understand the theoretical foundations underlying the analysis of EEG, Wi-Fi sensing, and other signals acquired from intelligent sensing systems, recognizing both the similarities and differences between visual and non-visual information processing. Students will acquire a comprehensive understanding of the major families of algorithms employed in modern Artificial Intelligence, including both classical Computer Vision approaches and hand-crafted techniques, as well as Machine Learning, Deep Learning, and state-of-the-art AI models. In particular, they will understand the operating principles of major neural architectures, including CNNs, ResNet, U-Net, Siamese Networks, Teacher–Student architectures, Graph Neural Networks, Vision Transformers, Foundation Models, Vision-Language Models, and Large Language Models, analyzing their strengths, limitations, and primary application domains. Particular attention will be devoted to the critical understanding of different computational paradigms, emphasizing that the selection of the most appropriate solution depends on the characteristics of the problem, data availability, computational complexity, hardware constraints, and application objectives. Students will therefore understand that classical algorithms, hand-crafted approaches, Machine Learning, Deep Learning, and Foundation Models should be regarded as complementary tools rather than universally applicable alternatives. Finally, students will acquire a comprehensive understanding of the principal application domains of modern Artificial Perception, recognizing how the techniques presented throughout the course can be employed in intelligent systems for robotics, unmanned aerial vehicles, autonomous rovers, underwater robots, satellite imagery, Human–Computer Interaction, biometrics, medicine, environmental monitoring, security, defense, smart cities, and many other scenarios in which the automatic perception, interpretation, and understanding of complex data represent key enabling technologies.
Applying Knowledge and Understanding
Upon successful completion of the course, students will be able to design, develop, implement, and evaluate intelligent Artificial Perception systems by critically applying the knowledge acquired to the analysis of images, videos, language, and heterogeneous signals. They will be capable of addressing complex Computer Vision and Artificial Intelligence problems through the design of complete processing pipelines, including data acquisition, preprocessing, data representation, feature extraction, computational paradigm selection, model training, experimental validation, and performance evaluation. Students will learn how to select the most appropriate computational methodology for a given problem, making informed decisions among classical Computer Vision and signal processing techniques, hand-crafted algorithms, Machine Learning approaches, Deep Learning architectures, and state-of-the-art Artificial Intelligence models. They will also be able to justify these choices according to the characteristics of the problem, the availability and quality of data, computational constraints, interpretability requirements, robustness, and application-specific objectives. Students will acquire the ability to design and implement intelligent systems addressing a broad range of Artificial Perception tasks, including image classification, object detection, segmentation, change detection, saliency detection, anomaly detection, tracking, action recognition, scene understanding, image captioning, visual reasoning, image retrieval, and advanced signal analysis. They will also be able to transfer similar methodologies to the analysis of EEG, Wi-Fi sensing, and other heterogeneous sensing modalities, understanding how different forms of information can be processed using common Artificial Intelligence paradigms. Particular emphasis will be placed on the development of intelligent systems for real-world applications, including autonomous robotics, humanoid robots, Unmanned Aerial Vehicles (UAVs), autonomous rovers, underwater robotic systems, satellite imagery and remote sensing, Human–Computer Interaction, biometrics, medicine, security, defense, and environmental and industrial monitoring. Students will therefore develop the ability to transfer the methodologies learned throughout the course to new application domains and emerging research challenges. A substantial part of the final assessment consists of the design and implementation of an original project, carried out individually or in small groups, through which students will demonstrate their ability to transform theoretical concepts into a complete, functional, and scientifically motivated Artificial Intelligence solution. The project also includes the preparation of a scientific-style technical report and a formal presentation of the obtained results, fostering experimental, methodological, and engineering skills that closely resemble those required in modern research and industrial environments.
Making Judgements
The course aims to develop students' ability to critically analyze and scientifically evaluate intelligent Artificial Perception systems. Upon successful completion of the course, students will be able to critically assess Artificial Intelligence algorithms, computational models, and system architectures in terms of accuracy, robustness, generalization capability, interpretability, computational complexity, scalability, and suitability for specific application domains. Students will understand that there is no universally optimal solution and that selecting the most appropriate methodology requires a careful evaluation of the problem itself, the quality and quantity of the available data, technological constraints, computational resources, and application requirements. They will therefore be able to compare different computational paradigms, identify their strengths and limitations, and provide scientifically grounded justifications for their design decisions. Particular emphasis will be devoted to developing the ability to independently consult the international scientific literature, understand the current state of the art, critically analyze scientific publications, and identify potential research directions and methodological improvements over existing approaches. Students will therefore acquire a research-oriented mindset, enabling them to distinguish between established methodologies, emerging paradigms, and experimental solutions while critically evaluating their scientific contribution. The course also promotes critical reflection on the methodological, experimental, and ethical aspects of Artificial Intelligence, addressing topics such as data quality, dataset bias, model robustness, interpretability, experimental reproducibility, system reliability, and the responsible use of Artificial Intelligence technologies across different application domains. These competencies will enable students to evaluate not only the quantitative performance of an intelligent system but also its scientific validity, technological sustainability, and practical applicability. Finally, the course encourages students to formulate original research ideas, identify open scientific challenges, and design innovative Artificial Intelligence solutions based on rigorous scientific reasoning. The overall objective is to cultivate independent, critical, and creative thinking, preparing students for both advanced research activities and high-level industrial innovation.
Communication Skills
Upon successful completion of the course, students will be able to communicate, in a clear, rigorous, and scientifically sound manner, problems, methodologies, computational models, experimental results, and design choices related to Artificial Intelligence-based Artificial Perception systems. They will be capable of describing the complete design process of an intelligent system, from problem formulation and data acquisition to model development, experimental validation, and performance evaluation, critically discussing the motivations behind the adopted solutions, their advantages, limitations, and potential future developments. Students will acquire the ability to present and compare different computational paradigms, explaining when classical Computer Vision and signal processing techniques, hand-crafted algorithms, Machine Learning approaches, Deep Learning architectures, or state-of-the-art Artificial Intelligence models represent the most appropriate solution for a specific problem. They will also be able to justify their design choices according to the characteristics of the problem, the available data, computational constraints, interpretability requirements, and application objectives. Particular attention will be devoted to communicating experimental results using appropriate evaluation metrics, quantitative comparisons with the state of the art, statistical analyses, graphical visualizations, and reproducible experimental protocols. Students will further develop the ability to critically interpret these results, distinguishing between meaningful improvements and variations arising from experimental conditions or methodological differences. These communication skills will be assessed primarily through the presentation and discussion of the project developed during the course. Students will be expected to clearly describe the addressed problem, illustrate the proposed system architecture, justify the adopted methodological choices, critically compare their work with the existing scientific literature, and discuss the obtained results, limitations, and possible future developments. Particular emphasis will also be placed on the preparation of a scientific-style technical report, organized according to the standards commonly adopted in the international Artificial Intelligence research community.
Learning Skills
Upon successful completion of the course, students will have acquired the methodological foundations necessary to independently explore emerging techniques, computational paradigms, and research directions in Artificial Intelligence for Artificial Perception, enabling them to continuously update their knowledge in a rapidly evolving scientific and technological landscape. Students will develop the ability to independently read, understand, and critically analyze scientific papers, technical documentation, and research contributions related to Computer Vision, signal processing, Machine Learning, Deep Learning, Foundation Models, Vision-Language Models, Large Language Models, and, more generally, modern Artificial Intelligence technologies. They will also acquire the capability to transfer established methodologies to new application domains, adapting computational models and algorithms to different sensing modalities and heterogeneous data sources. The course aims not only to provide specific technical knowledge, but also to develop a rigorous scientific methodology for addressing complex research and engineering problems. Students will therefore learn how to formulate Artificial Intelligence problems, identify appropriate data representations, critically select computational paradigms, design rigorous experimental protocols, objectively evaluate obtained results, and compare their solutions against the current state of the art. Particular emphasis will be placed on fostering an innovation-oriented mindset, in which continuous learning represents a fundamental requirement for coping with the rapid evolution of Artificial Intelligence technologies. Students will therefore develop the ability to integrate new algorithms, computational models, and emerging paradigms into their own knowledge framework while maintaining a rigorous, critical, and scientifically grounded approach to problem solving. Upon completion of the course, students will therefore possess the knowledge, methodological skills, and critical thinking necessary to autonomously pursue further education, scientific research, and professional development in Computer Vision, advanced signal analysis, Artificial Perception, and, more broadly, modern Artificial Intelligence.
|
