The Italian National PhD Program in Artificial Intelligence is made of 5 federated PhD courses that bring together 61 universities and research institutions. The 5 PhD courses share a common basis in the foundations and developments of AI, and each one has an area of specialisation in a strategic sector of AI application. 
The areas of specialisation are:  

  • Health and Life Sciences (Lead University: Università Campus Bio-Medico di Roma)
  • Industry 4.0 (Lead University: Politecnico di Torino)
  • Security and Cybersecurity (Lead University: Sapienza Università di Roma
  • Environment and Agricolture (Lead University: Università degli Studi di Napoli Federico II)
  • Society (Lead University: Università di Pisa)

The role of the Department

The University of Trento participates in the national PhD Program in Artificial Intelligence in the AI ​​for Society area, coordinated by the University of Pisa, thanks to the scientific excellence and relevance of the Department of Information Engineering and Computer Science in the field of  Artificial Intelligence. 
All details about the program are available at https://www.phd-ai.it/en/359-2/

Area AI for Society

The study of society and the complexity of social and economic phenomena has received a strong boost in the last ten years thanks to the AI and Data Science methods, powered by the social microscope of big data analytics and social mining through inter-disciplinary hybridization with the social and economic sciences. The combination of model-driven and data-driven approaches of data mining, machine learning and network science is progressively increasing the ability to observe, measure, model and predict complex socio-economic phenomena, such as human mobility and dynamics of cities, migrations and their economic determinants, the dimensions of community well-being, the formation and dynamics of opinions and online conversations, and the social impact of AI systems. This scientific line is interlaced with the Human-centric AI one, meaning the development of advanced forms of person-machine interaction capable of improving the quality of individual and collective decisions in delicate fields, from health to justice, to economic transactions, to risk assessment in various social and economic areas. The AI ​​for Society specialization area will focus on crucial issues such as explainable AI, AI for personal assistance, AI for social interaction, AI for social good, following an approach aimed at incorporating shared ethical values ​​in AI systems (ethics-by-design) and to achieve common goals, with a view to sustainability, diversity, respect for human dignity and autonomy, inclusiveness and social acceptability.

Call for admissions

The call for admissions to the "AI for Society" area, for a.y. 2023-2024 is now open!
Apply by August 7, 2023, h 13:00 CET athttps://dottorato.unipi.it/index.php/it/concorsi-d-ammissione-a-a-2023-2...

Specific co-financed Projects - cycle 39 - a.y. 2023-2024

The University of Trento is involved in cycle 39th of area "AI for Society" by co-financing two specific projects:

1. Operational Optimal Planning for Healthcare Coordination
The goal of this PhD scholarship will be to develop novel advanced planning and scheduling algorithms leveraging neuro-symbolic hybrid approaches to orchestrate and coordinate the activities within the healthcare system, by ensuring robustness and resilience to contingencies, accounting for multi-objective cost functions, and eventually providing explanations about the suggested solutions.
Contact: prof. Giovanni Iacca (giovanni.iacca [at] unitn.it) and prof. Marco Roveri (marco.roveri [at] unitn.it)

 2. Federated multi-targed domain adaptation
Federated learning methods enable us to train machine learning models on distributed user data while preserving its privacy. However, it is not always feasible to obtain high-quality supervisory signals from users,
especially for computer vision tasks. Unlike typical federated settings with labeled client data, this research will consider a more practical scenario where the distributed client data is unlabeled, and a centralized labeled dataset is available on the server. The research will also consider the server-client and inter-client domain shifts into account and pose a domain adaptation problem with one source (centralized server data) and multiple targets (distributed client data).
Contact: prof. Nicu Sebe (niculae.sebe [at] unitn.it)

Specific co-financed Projects - cycle 38 - a.y. 2022-2023

The University of Trento is involved in cycle XXXVIII of area "AI for Society" by co-financing two specific projects:

1. Tuning of music information retrieval models via evolutionary computing techniques.
The PhD will focus on the application of evolutionary computing methods for the optimization of machine learning models in different music information retrieval tasks. The successful candidate will design, implement and evaluate advanced techniques merging the domains of music information retrieval and evolutionary computation, in areas such as classification of genres, emotions, audio effects, and type of instrument in large datasets of musical signals. Both offline and real-time scenarios will be investigated.
Contact: prof. Luca Turchet (luca.turchet [at] unitn.it)

2. Towards hybrid human-machine learning and decision making.
The project will focus on the development of hybrid strategies combining human decision-makers and machine learning algorithms to improve the performance of the joint human-machine system. This challenging goal requires an interdisciplinary perspective, combining aspects of explainable AI, interactive machine learning, human-computer interaction, human decision-making and cognitive science. A relevant case study will be the development of hybrid strategies for effective public policy making.
Contact: prof. Andrea Passerini (andrea.passerini [at] unitn.it)

Specific co-financed Projects - cycle 37 - a.y. 2021-2022

The University of Trento is involved in cycle XXXVII of area "AI for Society" by co-financing three specific projects:

1. Human-cognition aware explainable AI 
Existing research in explainable artificial intelligence is mostly focused on either explainable by-design methods or post-hoc reverse engineering approaches. This research aims to broaden the scope of explainable AI bringing the human in the loop of the learning process itself. This requires on the one hand to develop interactive approaches, where the machine and the user engage in a dialogue trying to improve their mutual understanding, and on the other hand to develop forms of explainability that are aware of the limitations and specificities of human cognition. The research will be conducted as a collaboration between DISI and CIMEC and the candidate will be jointly supervised by faculty members of the two institutions.
Contact: prof. Andrea Passerini (andrea.passerini [at] unitn.it)

2. Deep learning models for multi-modal human behaviour analysis and synthesis
This PhD project has the ambition to explore the fusion of multiple modalities (e.g. video, audio, inertial sensors, etc.) and the design of novel cross-modal deep neural network architectures to study social behaviours, social interactions, and human activities. In addition, the project will also address the challenge of exploiting deep generative models such as Generative Adversarial Networks (GANs) or Variational Autoencoders (VAEs) to generate video sequences depicting realistic human behaviours in a variety of social settings.
Contact: prof. Niculae Sebe (niculae.sebe [at] unitn.it)

3. Social robotics for elderly assistance
The project will aim to develop novel approaches for improving the perceptual capabilities of humanoid robots in the context of social interactions with elderly patients. Specifically, the project will aim to design and implement novel deep learning architectures which enable the robot to analyze human behaviours in multi-modal multi-party interactions, with the ultimate goal of building self-aware robots which understand the level of acceptance from the user. The PhD student will investigate algorithms for people detection and tracking, group analysis and facial expression recognition. These activities will require the design of specialized deep networks which operate in a resource-constrained setting and which permits the robot to adapt its internal knowledge to dynamic environments.
Contact: prof. Elisa Ricci (e.ricci [at] unitn.it)


For further information on the call, candidates are invited to check the webpages available in the "Useful Links" box.