Artificial Intelligence (AI) is an umbrella term, covering a large and heterogeneous amount of disciplines. This course aims at providing an overview of the foundations of AI and of its main disciplines (e.g. problem solving, knowledge representation and reasoning, planning, uncertain knowledge, learning, perception, ...) in an organic way. Examples and exercises will be provided during the course.

The course aims at studying the fundamentals of machine learning, covering supervised and unsupervised learning methods and deep learning approaches. It includes application examples as well as laboratory exercises.

Natural Language Understanding is the fundamental component of artificial intelligence systems ( AIS ) that communicate with humans. AIS communicate directly with humans via the conversational interfaces of social robots. AIS may be able to read and comprehend vast amounts of human language data ( speech, text or multimedia ) and makes sense. In the first part of the course we will provide the students the basic knowledge about the natural language structure from the lexicon to the document level, formal models for representing the lexicon, the sentence and the discourse. We will present and provide the machine learning models to learn language structures from language corpora. In the last part of the course we will describe use cases of Natural Language Understanding in AIS. Students will be trained to train simple natural language understanding models for different use cases.

Students will acquire the ability to understand principles of cognitive neuroscience that are relevant for understanding current thinking in AI and Cognitive Computational Neuroscience. They will be able to consider application of neuroscience to AI, and conversely, potential applications of AI to neuroscience, taking into consideration both similarities and differences in the representations of the human and artificial systems.

The course provides the basic competences in the field of digital signal processing, with special attention to images and video sequences. Starting from the fundamentals of 1D signal analysis and processing, analog and numerical, in time and frequency domains, we then extend the concepts to the multi-dimensional case of signals in space. Then, we introduce the more important approaches for image filtering and extraction of image descriptors. These concepts are further extended to deal with motion pictures. Finally, the problem of image compression is introduced, focusing on most known techniques for image and video coding, as well as their standard implementations. The approach of the course is rather practical, with the explanation of theoretical concept followed by their translation into algorithmic terms.

The course objective is to educate professional figures fully aware of the complex impact of AI and robotics on our society. In this respect, the student will be offered basic information on the ethical and legal implication of AI. Specifically, the course will expose the most important motives for a truly human centred AI and to the bioethical principles that could underly its construction. At the same time, the course will present and discuss the main regulatory tools that could be applied to AI at different levels (national regulations, European regulations, international regulations). The students will be interactively involved through the discussion of realistic cases and through the presentation of papers.

The course objective is to provide competencies on: i) the formalisms to specify planning and scheduling problems for an agent taking into account his/her capabilities and the expressivity of the planning objectives to achieve; ii) the different techniques to solve the defined problems; iii) the tools implementing the analyzed modeling and solving techniques.

The course provides the foundations of automated reasoning (AR). The main logic formalisms and algorithms for automated reasoning in the main domains of interest (e.g., SAT, SMT) are presented and illustrated by means of application. Lab sessions are provided where students will use state-of-the art AR systems.

Robots that can talk or chat with humans are becoming pervasive in many industry domains. AIS systems using natural language can interact with human and operate in command-and-control, information retrieval or cooperative decision-support tasks. In this course we review the basic principles of human computer interaction, conversation linguistics, discourse analysis, computational Dialogue Models, dialogue system architectures and their evaluation. In the second part of the course we provide methodologies for the design of Conversational Agents, data-driven training, design tools and a project-based lab addressing real use cases.

The course offers a bird-eye view on the most important topics related to modern robotics, highlighting the main problems that need to be faced in order to make the robots operate correctly in the environment. Specifically, the course will cover the following topics: taxonomy of the different types of robots, physical modelling (direct and inverse kinematics), simulation of models, sensing and actuation solutions for perception and action, motion planning. The course will adopt a combination of simulation models and lab experiences to consolidate the knowledge transmitted during the theoretical lessons.

The complexity of multi-agent systems has led to the definition of different development methodologies, software architectures and programming languages ​​in which the concept of agent assumes a key role as well as the concept of object-oriented object-oriented design. The objective of the course is to examine and explore the credentials of agent-based approaches as paradigm for the development of autonomous systems. We will present analysis and design techniques, architectures, algorithms, and tools for the development of agents. In particular, the course will deal with the agent concept and the development of multi-agent systems, architectures and algorithms for agent software design, planning principles, modeling techniques and goal-oriented analysis, communication languages ​​and agent-oriented software engineering methodologies.

The course aims to provide the student with an in-depth overview on the methods of analysis and management of multimedia data. Starting from the basics of image and video processing, the course will focus on the problems of motion modeling and detection, motion tracking, and object recognition, both by using monocular and multi-view systems.

The course aims to provide the students with the knowledge necessary to deal with complex problems in the field of computer vision. In particular, the course will give the students the theoretical and practical notions related to the main methods and algorithms for analyzing visual data based on neural networks and deep learning. The laboratory activity will complement the theoretical part of the course and will focus on the use of the main deep learning and computer vision libraries.

The course aims to make the students aware of the most recent developments in the area of Computer Vision. To this end, the course’s teachers and tutors will carry out an open discussion and presentation on the most relevant scientific papers, highlighting the theoretical foundations and practical implications through ad hoc laboratory experiences. The final exam will revolve around a scientific paper chosen in agreement with the teachers; the students will present the advantages and disadvantages of the technique presented in the paper, touching on such issues as the replicability of the results and the datasets. The discussion will follow the pattern of a scientific peer-review and the goal of the exercise is to deepen the scientific competences of the students and to strengthen their soft skills at the same time.

This course aims at presenting advanced topics on machine learning and optimisation research and technology. It will cover some of the most promising directions of recent research. The course includes lab exercises and seminars on selected topics.

The course topic is the application of AI tools to the perception of autonomous agents, with a particular emphasis on the distributed solutions (distributed perception and estimation). Bayesian and Markovian approaches will be introduced, together with Maximum Likelihood estimators, for classification, localization and motion predictions of the different entities acting in the operative scenario. The theoretical topic will be consolidated through simulations and lab experiments.

This course focuses on control of robotic systems, with special attention to numerical optimal control and reinforcement learning. After reviewing the basic principles of robot modeling and numerical optimization, students will learn different control techniques, from the simplest and most well-known, to the most recent and advanced. Methods will be first studied in theory, and then implemented in simulation (with the Python language) to gain practical experience. Applications will span industrial manipulators, legged robots, flying robots and wheeled robots. After completing the course, students will be able to: 1. understand the working principles of several control algorithms for robotic systems, 2. choose the appropriate approach(es) to control a specific system for a given target application, 3. implement, tune, and test control algorithms with the Python language

The course will delve into robotic deliberation, meaning the robots' ability to receive/decide a mission and to refine it into an executable and detailed tactical plan that fulfills its goals. More often than not, the latter amunts to deciding a trajectory that needs to be followed, avoding obstacles and cooperating with humans and other robots The student will receive a comprehensive introduction to the most important motion planning techniques and will refine her/his understanding through challenging laboratory experiences.

In the course, we will provide the basis for the application of AI techniques to product and process innovation.

The course will provide the student with fundamental knowledge on Microeconomy and business organisation, with particular regard to the economy of information, networks, and innovation in the areas of information and communication technologies and information systems. A particular attention will be devoted to decision making and management, and to the main factors affecting the decision of the stakeholders in companies, networks and markets. These principles have a primary importance both in the management of companies and in the definition of public policies for the regulation of markets.

The course introduces: 1. the fundamental techncologies for radar and remote sensing data acquisition (multispectral, hyperspectral, synthetic aperture and lidars), and 2. The properties of the related big data acquired by satellites, aircrafts and drones. Then it presents the most important techniques for automatic extraction of semantic information, for semantic segmentation, for change detection and fusion of sensor data. The addressed techniques leverage a combination of statistical methods, machine learning and physical models.

This course provides a comprehensive overview of the core topics in cognitive neuroscience, including perception, attention, memory, language, concepts, spatial and social cognition. We will present both classical and recent research findings obtained with a variety of cognitive neuroscience methods (fMRI, EEG, MEG, TMS, behavior, neurological patients).The objective of the course is to provide the students with the following competences:- Knowledge of the main topics in cognitive neuroscience- Knowledge of the main experimental techniques and methods- Ability to define specific experimental design to answer experimental questions- Ability to critically understand experimental data and results- Knowledge of the main open questions in cognitive neuroscience

The course delves into the problem have to be addressed across the boundaries between meaning, persuasion, language understanding and cognitive social processes: why is so difficult to understand correctly, why is it difficult to lie. The course will illustrate how persuasion is generated and how this is related with stereotypes, with the natural difficulty in revisiting the knowledge acquired in the past, with the theory of negations and with how language modifies our behaviour.

The purpose of this course is to give the students the main concepts and basic methodologies for the study of language, divided according to the following thematic areas: phonetics, phonology, morphology, syntax, semantics, discourse. The main approach will be that of theoretical linguistics, in the generative framework. The course strives to stimulate an active and participative approach to learning and stimulate "problem solving". The students will have to use the skills acquired to solve various real Language problems, or do field work, using the languages spoken in class as a training field, in order to acquire a concrete understanding of what it means to carry out research in linguistics. Those students who already have a theoretical linguistic background will have an opportunity to strengthen their knowledge by doing more advanced exercises and activities, which will be planned together with the teacher.

"The goal of this course is to provide a good understanding of vehicle dynamics and control systems through a combination of classroom-based theory sessions, hands-on computer simulation and analysis of real data and example of recent advanced automotive applications. The aim of the course is teaching the student how to model vehicle systems and sub-systems in order to study/optimise vehicle dynamics and design control systems. Finally the course will provide the basic knowledge of intelligent vehicle architectures and the main techniques for trajectory plannings."

The course introduces some computational techniques for the study of:- Ordinary Differential Equations- Laplace and Z transform- Constrained maxima and minima- Optimal Control problems.

The objectives of the course are threefold: 1. Introducing the students to modern design of mechatronic systems, following the life cycle approach in its different phases: requirement definition, specification, conceptual design and simulation tools; 2. Providing the knowledge and the abilities required for the simulation of multibody and multidomain systems, which can be used for concept evaluation and for the generation of open loop models usable in control design, 3. Evaluating the accuracy of the model and the final result against the initial specification.

Provide the essential tools for the analysis of dynamical systems in the continuous and discrete time domain, from linear time-invariant systems, to time-varying ones and up to nonlinear hybrid systems.

The course will be on basic neural anatomy and on the application to neurocognitive sciences of such neuroimaging techniques as functional and structural magnetic resonance imaging, transcranial magnetic stimulation, Magnetoencephalography, and EEG. At the end of the course, the students will have the following abilities related to brain imaging: 1. Defining the right imaging method for each cortical source, 2. Identifying the best equipment to use, 3. Explaining advantages and liitations of each technique, 4. describing the best methods for data analysis, 5. Describing and interpreting neuroimaging applications to neuroscientific research.

The course deals with themes, problems and methods related to the current research on formal semantics of the natural language. The course will also highlight the main differences between model theoretic semantics and approaches such as embodied mental representations and corpus-based distributional semantics. The course is a natural follow-up of the course “Introduction to human language” and complements its topics. At the end of the course, the students will have the following competences: 1 basic understanding of formal logic (propositional and predicate calculus), 2. computing the truth of logic formulae using truth table and refutation trees, 3. Evaluating the semantic value of a formula in a formal model, 4. Translating natural language quantified statements (in English or Italian) into extended predicate logic, 5. Using lambda calculus to obtain the truth value of complex expression in natural language from the truth value of its constituents, 6. Formulating the problem of the semantic coverage of advanced natural language phenomena such as ambiguity in quantification, event semantics, the correct expression of plurality, the distinction between mass and count terms and between distributive, collective, and cumulative predicates 7. Ability to read a scientific paper through formal semantics extracting at least the most salient points.

This is an introductory course to data mining and massive data analytics. In this course we will talk about some basic data mining techniques, such as association rules, sequential patterns, clustering, and classification. We will discuss different flavors of these techniques, and comment on their strong and weak points. Finally, we will also talk about the new trends in data analysis, and explore some emerging applications in this area, such as mining of streaming data (that is, data that is continuously generated).

The course is intended to provide the fundamentals of High Performance Computing software for High-end Data Science. The course will include both theoretical and practical aspects related to simulation-centric and data-centric paradigms; it will present convergent aspects of HPC software ecosystems and data frameworks at a large-scale in relevant scientific disciplines.

The course aims to extend students’ knowledge of grammatical, lexical and textual features of written academic English in a scientific context, and to develop students’ fluency in speaking English. An active approach is used, with students producing written texts and then correcting them individually and as a group.