Alessandro Farinelli is full professor at the University of Verona, Department of Computer Science.
His research interests focus on the development of innovative methodologies for Artificial Intelligence systems with applications in the field of robotics. In particular, his main research topics focus on:
-- coordination of multi-agent systems
-- distributed optimization
-- reinforcement learning
-- data analysis for cyber-physical systems.
Alessandro Farinelli has been scientific responsible for several national and international research projects focused on issues related to the development of Artificial Intelligence systems. His scientific contributions target mainly international journals in the area of Artificial Intelligence (e.g., Artificial Intelligence and Journal of Artificial Intelligence Research) and intelligent robotic systems (Autonomous Robots and Robotics and Autonomous Systems). The major conferences he contributes to, both as a speaker and organizer, include the International Conference on Autonomous Agents and Multi-Agent Systems (AAMAS), the International Joint Conference on Artificial Intelligence (IJCAI), and the International Conference on Intelligent Robots and Systems (IROS).
Modules running in the period selected: 48.
Click on the module to see the timetable and course details.
Di seguito sono elencati gli eventi e gli insegnamenti di Terza Missione collegati al docente:
Topic | Description | Research area |
---|---|---|
Intelligent agents | Design and development of autonomous entities that can sense, model and interact with the environment in which they operate. These area focuses on the interaction and integration of solution technques for several research topics such as automated planning and reasoning, reinforcement learning, statistical learning and reasoning in face of uncertainty. |
Artificial Intelligence
Distributed artificial intelligence |
AI & robotics | Application of AI techniques to increase the autonomy level of robotic systems. This includes the adaptation of algorithms for autonomous planning and reinforcement learning to: i) handle the cyber-physical constraints imposed by robots operating in partially observable and uncertain scenarios; ii) guarantee the reliability and robustness of robotic systems that operate in open environments (e.g., interacting with humans and other robotic systems); iii) facilitate the use of robotic systems in realistic application by proposing novel paradigms of interaction with users (e.g., train a robot to execute a task rather than specify a control program). |
Artificial Intelligence
Planning and scheduling |
Unsupervised learning | Is an approach where models are trained on unlabeled data, with the goal of identifying hidden patterns or structures within the data without predefined labels. It is commonly used for tasks like clustering, dimensionality reduction, and anomaly detection. Open research in unsupervised learning focuses on improving the ability to discover meaningful structures in complex, high-dimensional datasets, often with limited prior knowledge. Key challenges include developing more effective clustering algorithms, improving the interpretability of models that uncover latent structures, and handling high levels of noise or sparsity in data. Additionally, there is ongoing work to bridge the gap between unsupervised learning and other paradigms, such as semi-supervised, self-supervised or contrastive learning, and to enhance the robustness of unsupervised models in real-world applications. |
Artificial Intelligence
Machine learning |
Reinforcement learning | Reinforcement Learning (RL) is a learning paradigm where agents to learn how to take a sequence of decisions through interactions with their environment. RL trains a model by considering a reward signal that is associated with the actions performed in the environment (high reward for good actions and the opposite). The model aims at optimizing the expected accumulated reward over time. RL is very intersting for practical applications (e.g., robotics, recommender systems) because it requires minimal specifications from the user and it can adapt to unpredicatble changes in the enrvironment. Main challenges relates to devising safe policies for the agents, e.g., learning while avoiding catastrophic falures (safe reinforcement learning and offline reinforcement learning), to properly evaluate the quality of a trained system, e.g., how can we guarantee that the agent will behave properly in unseen situations, and to improve sample efficiency, e.g., model-based reinforcement learning. |
Artificial Intelligence
Machine learning |
Supervised learning | Is an approach where models are trained on labeled data to learn a mapping from inputs to outputs, enabling them to predict correct labels for new, unseen data. While widely used for tasks like classification, regression, and time series forecasting, open research in this field addresses several challenges. Key questions include how to make models more robust to label noise and inconsistencies, improve sample efficiency to reduce the need for large labeled datasets, and enable effective transfer learning across different tasks and domains with limited labeled data. Additionally, addressing issues of fairness and bias in supervised models, as well as improving scalability to handle large datasets without compromising performance, and attention/transformer-based approaches remain active areas of exploration. |
Artificial Intelligence
Machine learning |
Deep learning | Focuses on training neural networks with multiple layers to automatically learn patterns and representations from large amounts of data. Using architectures such as convolutional neural networks (CNNs) for images, recurrent neural networks (RNNs) for sequential data, and transformers for diverse tasks, deep learning excels at complex tasks like image recognition, natural language processing, speech recognition, reinforcement learning, time series analysis, and autonomous driving. |
Artificial Intelligence
Machine learning |
Explainable artificial intelligence | The goal of explainable AI (XAI) is to i) explain black-box models; ii) develop AI models which are interpretable by construction. For instance, this involves causal analysis and discovery, logical models of agency (with logic programming) and logical machine learning (with inductive logic programming). XAI helps characterize model accuracy, fairness, transparency and outcomes in AI-powered decision making; moreover this field focuses on methods for improving model and decision interpretation using statistical and graphical tools. |
Artificial Intelligence
Machine learning |
Planning under uncertainty | Planning under uncertainty focuses on sequential decision-making in uncertain environments, namely, situations with imperfect information. (Partially Observable) Markov Decision Processes are used to represent these contexts. The goal of planning under uncertainty is to generate optimal policies for these problems, namely, functions able to suggest optimal actions in situations faced by the agent. The main challenges concern dealing with large problems (scalability), acquiring new knowledge about the environment (adaptability), preventing undesirable behaviors (safety), safe policy improvement (robustness), interacting with humans (human-in-the-loop), supporting human understanding (explainability), bridging planning and reinforcement learning (model-based RL), bridging symbolic and probabilistic/data-driven planning. Among the most recent approaches to tackle these challenges, online methods based on Monte Carlo Tree Search have achieved strong results in the last years in both strategic games (e.g., board games such as Go) and real-world applications (e.g., robotics, cyber-physical systems, and decision support systems). |
Artificial Intelligence
Planning and scheduling |
Multi-agent planning | Multiagent planning deals with planning approaches applied to multi-agent systems. The main goal of these techniques is to generate solutions for sequential decision making that promote synergy among multiple autonomous agents to achieve collective goals. Among the main topic of this field there are decentralized optimization, multiagent path planning, multiagent learning, cooperation and coordination. Important tools in this fiels are, for instance, coordination graphs that are used in recent cooperative multi-agent planning and reinforcement learning (MARL) algorithms where coordination between agents is essential to accomplish the task. Coordination graphs allow to represent how agents can coordinate using some communication via message passing. Applications of multiagent planning span over a wide set of domains including autonomous driving, logistic (e.g., fleet of autonomous robots), environmental monitoring (fleet of mobile drones for data acquisition). |
Artificial Intelligence
Planning and scheduling |
Neurosymbolic planning | Neurosymbolic AI focuses on combining standard data-driven AI (e.g., reinforcement learning) with symbolic approaches (e.g., logic programming and inductive logic programming), in order to enhance the explainability of AI systems (e.g., autonomous agents), their efficacy in human-robot interaction, and foster incremental knowledge acquisition and generalization in planning. |
Artificial Intelligence
Planning and scheduling |
Multi agent systems | Design and development of multiagent systems, where intelligent agents can interact among them, with the environment and with humans. This area focuses on the interaction and integration of solution techniques related to multiagent planning, statistical learning, multi-agent reinforcement learning and game theory. |
Artificial Intelligence
Distributed artificial intelligence |
Office | Collegial Body |
---|---|
member | Faculty Board of PhD in Computer Science - Department Computer Science |
member | Computer Science Teaching Committee - Department Computer Science |
member | Information Engineering Teaching Committee - Department Department of Engineering for Innovation Medicine |
member | Computer Science Department Council - Department Computer Science |
department director | Academic Senate |
******** CSS e script comuni siti DOL - frase 9957 ********p>