Reserved topic scholarships - 2022 - 2nd call | Doctoral Program - Information Engineering and Computer Science

Reserved topic scholarships - 2022 - 2nd call

Department of Information Engineering and Computer Science

A1 - Adaptive Multimodal conversational agents (1 grant)

Thanks to the research on multi-task learning and on transferability, the current trend in natural language understanding is to build universal models: models pre-trained on several tasks and fine-tuned on a downstream task to which the acquired conceptual knowledge and skills are transferred. These steps are very important contributions to AI, and in particular to the development of conversational systems, but universal models do not model human intelligence diversity. We believe it is important to work with conversational models which have different prior experiences and use different conversational strategies. With the progress on modelling vision and language interaction, current dialogue models are able to deal with visually grounded language. However, they are not able to adapt their language use to the interlocutor, hence they miss an essential ability to be engaged in effective communication. ADAPT aims to address this problem.
To accomplish our overall goal, the research agenda of ADAPT capitalises on recent advances in  the development of multimodal  models; on the availability of multimodal datasets for referential  visual games with textual interactions among players; and on methods developed recently within the research on emergent communication in referential  games.  To these active and promising fields, we will add a fundamental missing component, namely the view about how humans take community mutual knowledge into account and manage to adapt to each other through conversations. 
The scholarship is supported by Amazon Alexa and the work will be run in collaboration with their research group.

Contact: Raffaella Bernardi raffaella.bernardi [at]

A2 - Distributed machine learning for smart buildings (project HE EIC SUSTAIN, CUP: E63C22001550006) (1 grant)

This project will focus on the design and development of lightweight machine learning models for sensor systems to be used in smart buildings. We will study different distributed learning approaches, based on consensus and other forms of aggregation of the outputs of the node-local models, in order to achieve a global goal at the network level. Federated and split learning approaches, also including probabilistic learning and evolutionary optimization, will be considered to ensure a flexible architecture capable of guaranteeing a proper separation of concerns and data protection/privacy. The proposed research will lie at the intersection of the following topics and can be structured considering the profile and interests of the candidate: Machine Learning (Tiny Machine Learning), Evolutionary Computation, Distributed Systems and Algorithms, Embedded Systems (Low-Power Computing), Sensor Networks (Low Power Communications and Networking). Previous experience in at least one of these topics is considered a plus.

Contact: Giovanni Iacca giovanni.iacca [at] 

A3 - Efficient machine learning for embedded systems (project HE EIC SUSTAIN CUP: E63C22001550006)  (1 grant)

This project will focus on the design and development of lightweight machine learning models for sensor systems to be used in smart buildings. We will investigate lightweight black-box (e.g., neural networks), white-box (e.g., decision trees) models, and combinations thereof. Moreover, we will use Neural Architecture Search and evolutionary algorithms to derive optimally designed models that take into account possible computational and energy constraints on the nodes of the distributed system. We will also study the explainability of those models. The proposed research will lie at the intersection of the following topics and can be structured considering the profile and interests of the candidate: Machine Learning (Tiny Machine Learning), Evolutionary Computation, Distributed Systems and Algorithms, Embedded Systems (Low-Power Computing), Sensor Networks (Low Power Communications and Networking). Previous experience in at least one of these topics is considered a plus.

Contact: Giovanni Iacca giovanni.iacca [at] 

A4 - Self-supervised incremental and adaptive learning (project MUR PRIN 2020, CUP E63C22000390001- Prot. 2020TA3K9N); (project UE H2020 "SPRING", CUP n. E64I19002980006)  (1 grant)

While modern AI systems exhibit astonishing performance reaching human-level capabilities in different applications, they are still severely lacking human abilities of learning incrementally and continuously, as well as quickly generalizing from few examples and adapting to new and unseen environments. In this PhD project novel theoretical and computational frameworks to build deep learning models that are able to adapt when data from new domains and new semantic categories are made available will be devised, with special emphasis on designing models which are able to learn without having access to data annotations. This grant is funded by the project MUR PRIN 2020 - LEGO.AI: LEarning the Geometry of knOwledge in AI systems, CUP E63C22000390001- Prot. 2020TA3K9N and the EU PROJECT SPRING.

Contact: Elisa Ricci e.ricci [at] 

A12 - Climate data integration for early warning and impact analysis of drought using machine learning and deep learning techniques (progetto HORIZON EUROPE - INTERTWIN -CUP: E63C22001530006) (1 grant)

Early warning systems of hydrometeorological extremes, e.g. floods and droughts, are becoming increasingly crucial due to increased population settlement in vulnerable areas, particularly prone to flooding or water shortage, and the projected increase of weather extremes due to climate change. 
In particular, drought can cause numerous and severe impacts on natural and socio-economic systems and its forecasting is still challenging due the complex nature of the phenomenon which has no unique definition, occurs over different temporal and spatial scales, is determined by different causes and whose effects have a delayed onset in relation to the start of the event. Since drought development is highly influenced by local climate conditions and their interactions with surface properties and processes, the availability of skillful sub-seasonal to seasonal predictions of potential climate drivers is an essential element for improving the drought forecasting as well as the management of its impacts at local and regional scale. Most decision making processes in a wide range of sectors, from agriculture to disaster risk reduction, are related in fact to this time range. While research community is constantly increasing the capabilities of operational physics-based forecasting models, machine learning and deep learning methods have also shown high potential to advance sub-seasonal to seasonal climate forecasting.
The PhD project will focus on the development of machine learning and deep learning methods to provide tailored sub-seasonal to seasonal forecasts of main climate variables together supporting the characterization and prediction of drought conditions across the Alpine region together with the estimation of uncertainty. The research activity will include the implementation and testing of different techniques, from support vector machine to convolutional neural networks, by using state-of-the-art data and the assessment of the most relevant predictors. The results will provide valuable scientific contributions towards the integration of emerging machine and deep learning for advancing sub-seasonal to seasonal climate systems.
The findings will be a key component of an integrated data-driven framework fusing multiple data sources for the prediction of drought impacts and supporting the implementation of early warning systems of drought at local and regional scale. Attributing climate change elements to drought occurrence will be an additional and crucial step to increase the understanding of key climate drivers. It will improve the quality and reliability of the forecasting system as well as the understanding of future evolution of drought in the Alps.

Contact: Sandro Luigi Fiore sandro.fiore [at]

B1 - Formal framework for Trustable IoT applications (1 grant)

This PhD aims at defining a formal specification framework for the design and deployment of trusted IoT applications considering all the different facets and by attaching to the IoT applications a “certification manifest”. The study will consist in formally defining the concepts of safety and assurance for IoT applications leveraging on temporal logic specifications complemented with security concepts, and adaptability of existing frameworks and verification tools to address the identified verification problems. The theoretical framework will be complemented with implementation activities to develop/adapt different verification tools for the different verification activities. The work will also consider i) studies aiming at adapting existing translation validation techniques to ensure the verification of safety and security policies for the considered IoT application when deployed in a specific trusted execution environment leveraging SMT techniques; ii) AI techniques to reconstruct from the binary code high-level functions and control flow graph to enable the verification of safety and security properties to assure security policies are not violated by updates/patches.

Contact: Marco Roveri  marco.roveri [at]

B2 - Resilient AI-Based Self-Programming (project MUR PRIN 2020 - CUP E63C22000400001; Prot. 20203FFYLK) (1 grant)

This PhD aims at developing a core theory and algorithms for resilient self-programming, i.e, to define mechanisms to enable agents to act in an informed and intelligent way in their environment, by changing autonomously the way they behave as a consequence of the information they acquire from the external world and exchange with the humans operating therein. The study will focus on extending existing approaches to AI Planning and LTL/LTLf synthesis by i) enriching
the computed strategies with fault-tolerant capabilities, ii) considering several models at synthesis and execution time leveraging the most appropriate model depending on the observed contingency at execution time; iii) integrating reinforcement and model learning to enable for determining tolerant strategies that work in a reference model plus variations. The theoretical framework is complemented by the realization of prototype supporting tools as well as practical applications in selected realistic scenarios. This grant is funded by the project MUR PRIN 2020 - RIPER - Resilient AI-Based Self-Programming and Strategic Reasoning - CUP E63C22000400001.

Contact: Marco Roveri marco.roveri [at]

B3 - Security firmware  for modern architectures of embedded systems (1 grant)

Modern micro architectures include several security mechanisms and features that are not always exploited by current software layers. This PhD focuses on the study and the design of novel security services that leverage on these mechanisms typically implemented in hardware. Thus the research activity includes the hardware and software co-design of new security services implemented in firmware. These designs will be realized and validated by means of experimental analysis.  These security services typically run inside a  trusted execution environment (TEE). Scope of the PhD is also the investigation of the security of existing TEEs in order to understand and mitigate the root causes of an increasing number of vulnerabilities that have been recently reported.

Contact: Bruno Crispo  bruno.crispo [at]

C2 - Innovative Education in Computer Science (1 grant)

This PhD research grant has the objective of designing, executing and evaluating highly qualified educational initiatives for Computer Science - and more in general, Engineering students – while exploring the following research questions:
(1) How can we create technical education that is resilient to technological obsolescence?
(2) What non-technical skills can technical education help develop, and how can they be taught?
(3) How can technical education effectively embed an orientation towards the social good, e.g., creating positive environmental and social externalities?
Candidates will work on developing innovative teaching based on active learning, also using pedagogical methods such as challenge-based learning. All activities will be contextualised in the frame of many European collaborations active within the research group. Ideal candidates will have (or will aim to develop) an interdisciplinary profile, combining a computer science core expertise with sound grounding on education theory.

Contact: Maurizio Marchese  maurizio.marchese [at]

C4 - HPC-based and FAIR-enabled big data infrastructure for climate change research at extreme-scale (project PNRR MS4 C2 CN0000013 – CUP E63C22000970007) [additional reserved topic scholarship]

The Department of Information Engineering and Computer Science of the University of Trento offers a PhD scholarship for outstanding students who are willing to advance the
software and data infrastructure needed for next generation Earth System Models (ESMs) workflows within the “Earth & Climate” Spoke of the recently established ICSC - Italian National Center for High Performance Computing, Big Data and Quantum Computing.
The activity will target challenges at extreme-scale regarding the development of a scalable HPC-based and FAIR-enabled digital infrastructure able to integrate into the same
environment tools, ESM components, and applications, but also data and workflows, to support climate scientists in terms of handling, processing and analysis of very large
volumes of climate data. The position will address data and computational challenges (including end-to-end aspects such as workflow and provenance management) at extreme- scale at the intersection of HPC, Big Data and Cloud Computing.
The activity will benefit from and will be strongly integrated with the resources and software platform made available by the ICSC Italian National Center.

Contact: Sandro Luigi Fiore  sandro.fiore [at]

D1 - Machine learning and artificial intelligence for planetary radars (project ASI JUICE RIME - CUP F65F21000950005) (1 grant)

Artificial intelligence methodologies and in particular deep learning techniques are gaining a large popularity in many  areas of computer vision and signal processing given their capabilities to outperfom traditional methods. In this context, it is of fundamental importance to define new AI methodologies that can support planetary missions and the related signal/image analysis tasks to enhance the capability to automatically extract information from the data. The research activities of this grant are related to planetary radar sounders, which are instruments for the study of the subsurface of the planets and the moons of the Solar system. These radars operate from satellite and acquire data related to the subsurface of celestial bodies that can provide groundbreaking science results. The activity will be focused on the development of methodologies based on AI and machine learning  for either supporting the design of the radar systems or the automatic analysis (semantic segmentation, classification, change detection) of the images (radargrams) acquired by the radar sounders.
The research will be developed at the Remote Sensing Laboratory ( and will be related to the Radar for Icy Moon Exploration (RIME) on board of the JUpiter ICy moons Explorer (JUICE) of the European Space Agency (see for more details on the mission) and the Sub-surface Radar Sounder on board the EnVision mission to Venus of the European Space Agency (see for more details on the mission).

Contact: Lorenzo Bruzzone lorenzo.bruzzone [at]

D2 - Machine learning and artificial intelligence for the automatic analysis of optical remote sensing images (1 grant)

Earth Observation satellites play a crucial role in a large number of application domains (e.g., precision farming, forestry, analysis of urban areas, climate change). The automatic extraction of the information from these data is crucial for their exploitation in real operational services. In this context, even if artificial intelligence and machine learning have been widely used in the past years, there are still many methodological challenges that should be addressed for a proper information extraction from satellite images. The research activities of this grant are related to the development of novel methodologies based on deep learning for the automatic classification of satellite images and/or the automatic change detection in multitemporal images. The research will be focused on the analysis of multispectral and hyperspectral images acquired by last generation satellites and the related application to real world scenarios. Specific attention will be devoted to the automatic analysis of images acquired by the very recent PRISMA hyperspectral satellite of the Italian Space Agency.
Research will be developed at the Remote Sensing Laboratory ( and will be linked to the project activities in progress on the aforementioned topics in the laboratory.

Contact: Lorenzo Bruzzone lorenzo.bruzzone [at]

Department of Information Engineering and Computer Science, Q@TN Project

C1 - QVNSIM - Scalable algorithms for quantum simulation on Multi-GPU systems and hybrid programming models (1 grant)

Quantum computing will propel a new opportunity in many areas of science.
The performance of quantum systems evolves. Since the technology is still not sufficiently mature to tackle practical problems, only by simulating tomorrow’s quantum computers on today’s classical systems researchers can design quantum
algorithms more quickly and at scales not otherwise possible.
The project aims to establish new state-of-the-art scalable algorithms for simulating hybrid quantum systems (QVN) on parallel systems. The framework will also include the definition of error models to validate hybrid computing models for graph algorithms.
The activity aims to explore three different directions and targets the following objectives:
1. Techniques for simulating both quantum and QVN systems and Quantum Processing Unit;
2. Models for understanding the reliability of quantum circuits against radiation-induced transient faults;
3. Definition of a programming model for graph analytics for hybrid QVN systems.

Contact: Flavio Vella flavio.vella [at]

Department of Information Engineering and Computer Science, Q@TN Project and 12thLevel Pty Ltd

A11 - Solving Logistic Problems using NISQ devices   (1 grant)

One of the most interesting challenges in modern quantum computing is how to handle the intrinsic noise of quantum devices. This problem is at the basis of what is known as the "noisy intermediate-scale quantum" (NISQ) era. Currently, there are several approaches to reduce noise in quantum hardware devices, such as quantum error correction techniques and classical post-processing methods. In this project, rather than attempting to control/reduce noise, we will explore the possibility of using noise in a meaningful way. The idea is to investigate how we can develop metaheuristic approaches using the intrinsic noise of NISQ devices. We will first try to characterise a model of the noise, and then apply this model to one or more (classical) metaheuristics. The resulting algorithms can be seen as an instance of quantum memetic computing, i.e., a form of hybrid algorithms that combine both classical and quantum computations to perform both local and global optimisation. We will then validate the proposed approach on industrially relevant scheduling and vehicle routing problems, on which we will perform a comparison against classical algorithms implemented in commercial MILP solvers. This project is co-funded by 12thLevel and will be conducted in collaboration with the University of Newcastle, Australia.

Contact: Giovanni Iacca giovanni.iacca [at] 

EURAC Research - Accademia Europea di Bolzano

D3 - Snow cover detection and glacier mass balance estimation with machine learning methods and multi-source satellite data (1 grant)

Changes in glacier area, elevation and mass are major indicators for climate change and are identified as Essential Climate Variables” by the World Meteorological Organization. This PhD project will focus on the development of machine learning methods for snow cover detection over glacierized areas and glacier mass balance estimation, exploiting different sources of satellite data. These will include high and medium resolution multi-spectral images (e.g. Sentinel-2 and Sentinel-3) as well as synthetic aperture radar data (e.g. Sentinel-1). The methodology will be designed and tested over in-situ monitored glaciers in Norway, Svalbard and European Alps. Then, the transferability to glacierized regions with less ground data available will be tested. The project will allow to build and automatically update a consistent time-series of glacier surface mass balance and area change. These are highly valuable data for the hydropower industry, governmental agencies and the research community, e.g. to improve runoff forecast for enhanced water management (e.g. drinking water, hydropower, agriculture) and to increase the knowledge about glaciers as climate variable, their fading in a warming climate and their contribution to global sea-level rise.

Contact: Mattia Cellegari mattia.callegari [at]; Lorenzo Bruzzone lorenzo.bruzzone [at]

Fondazione Bruno Kessler (FBK)

A5 - AI-based 3D inspection for industrial quality control (1 grant)

Machine and deep learning methods are entering also the industrial sector to automatise 3D monitoring and analysis tasks. The research should investigate the use of AI-based methods to boost photogrammetric 3D inspections for industrial quality control operations. Innovative and advanced AI-based solutions should be developed in order to inspect non-collaborative surfaces (reflective, transparent, etc.) and derive precise 3D results useful for quality control.

Contact: Fabio Remondino remondino [at]

A6 - Application-oriented Speech Translation (1 grant)

The need to translate audio input from one language into text in a target language has dramatically increased in the last few years with the growth of audiovisual content freely available on the Web. Current speech translation (ST) systems are now required to be flexible and robust enough to operate in different application scenarios. On one side, the industry calls for key features like real-time processing, domain adaptability, extended language coverage, and the capability to adhere to application-specific constraints. On the other side, society calls for new efforts towards inclusiveness with respect to specific categories and groups (e.g. gender-sensitivity, customization to the needs of impaired users). Both industry and society face the orthogonal challenges posed by the variability of audio conditions (e.g. background noise, strong speakers’ accent, overlapping speakers). The objective of this Ph.D. is to make ST flexible and robust to these and other factors.

Contact: Matteo Negri negri [at]

A7 - Deep learning for vision-based scene understanding (1 grant)

Supervised learning is a popular mechanism to teach machines vision-based tasks and skills. However, human supervision is a bottleneck for building generic machines that can operate across different contexts, environments and applications. Ideally, machines should develop their own effective and possibly creative strategies for using the sensed data and their experience to continually learn without humans at their side. The research activities related to this PhD position will focus on building novel deep learning-based vision algorithms to teach machines to seamlessly understand environments through 2D or 3D perception.

Contact: Stefano Messelodi messelod [at]

A8 - Self-configuring resource-aware AI-based speech processing (1 grant)

The goal of the thesis is to develop AI models for speech processing which are aware of the computational resources and of the application requirements and are capable of dynamically adapting in order to meet such limitations. This entails not only the search for a trade-off between resources and inference performance but also the possibility to dynamically exploit additional computational resources, eventually expanding the model. The project will address both training and inference phases, starting from state of the art supervised techniques as model compression, neural architecture search, distillation and continual learning and pushing them towards continuous and unsupervised solutions.

Contact: Alessio Brutti  brutti [at]

A9 - TinyAI for energy-efficient smart sensing in distributed IoT (1 grant)

Machine learning and deep neural networks have been extensively and successfully used to process multimodal data (e.g., audio, video, environmental data) on powerful computers. At the same time, several challenges remain open to move AI onto low-consuming, resource-constrained devices (e.g., end nodes in an IoT). Recently TinyML approaches are emerging to distribute the intelligence to the far edge of the edge-to-cloud continuum. Exciting research scenarios emerge, spanning from tiny deep learning solutions for inference on resource-constrained platforms based on distillation, quantization, or neural architecture search going up to combing software techniques with novel, innovative hardware supporting TinyML. The complexity grows if we consider moving learning to the edge in order to benefit from the opportunities offered by connected, distributed devices. Motivated by these scenarios, the research aims (i) to define novel hardware/software approaches to optimize AI at the very edge on energy-efficient embedded devices, in particular for audio processing and/or computer vision, but not only; (ii) to explore the potential of distributing and fusing intelligence from heterogeneous nodes of an IoT (iii) to demonstrate the advantages of the investigated approaches in real-world application scenarios, such as those of smart cities.

Contact:  Elisabetta Farella efarella [at]

A10 - Artificial Intelligence for the Earth Systems (1 grant)

Climate change and its impact on countless sectors of society has enormously increased the demand for a comprehensive, robust, timely and reliable climate data analysis that provides support to the adaptation and mitigation policies. Earth System Models (ESMs) that faithfully simulate the cycle of the different components of the Earth System (atmosphere, hydrosphere, cryosphere, biosphere) are the key to address the complex challenges the society is facing, and their development requires expertise at the border between physics and computer science. During the PhD the student will be guided in exploring and applying Artificial Intelligence methods for the parametrization of physical processes, leveraging explainable AI and physics-informed machine learning and HPC-enabled large scale data understanding and processing, which try to blend machine learning with physical knowledge to achieve solutions that are physically more consistent.
The activity will be carried out in collaboration with Fondazione Bruno Kessler and within the activities of Earth & Climate Spoke of the National Center for High-Performance Computing (HPC). Candidates familiar with physical process simulations are welcome, and basic knowledge of Machine Learning/Deep Learning is recommended.

Contact: Grabriele Franch franch [at] / Marco Cristoforetti marco.cristoforetti [at]

B4 - Combining automated planning and deep learning for automatic adaptation (1 grant)

Automated planning is successfully used in some application areas for the synthesis of plans to control complex systems. Despite significant progress in the literature, scalability is still a major problem that hinders adoption of planning in a wider range of domains.
In this PhD research in the area of integrative AI, the candidate will study methods for combining modern deep learning approaches with symbolic AI for the automatic adaptation of planning tools. In particular, he/she will develop algorithms to automatically specialize planners on specific domains to improve on scalability by exploiting the characteristics of the target domain extracted automatically by means of machine learning.

Contact: Andrea Micheli amicheli [at]

C3 - Analysis and modeling of online communication networks (1 grant)

In the last decade, Social Media Platforms have become our main communication hub, encompassing both our personal and our public life. As online social networks have established themselves as important sources or information, they have rapidly changed the landscape of the news media ecosystem on a global scale. News is no longer exclusively broadcast by established sources. Within the participatory environment of these platforms, new opinion leader often actively creates and disseminate news without the restrictions posed upon classical media channels, and often reach large audiences. In this PhD project, we want to investigate how the structural and functional characteristics of online communication networks influence the circulation of news with a focus on disinformation and more broadly junk news. We further want to investigate how such unreliable information diffuses differently across heterogeneously formed communities and characterize the behavior surrounding their reception. We will use methodologies coming from network science, data science and complexity science, integrating the insight about the role taken users that can be obtained by the analysis of the communication network and the associated spreading dynamics with insight about the stance of those users that be automatically extracted using NLP methods.

Contact: Riccardo Gallotti rgallotti [at] 

D4 - Analysis of long and dense remote sensing image time series (1 grant)

In the context of the green deal transition and climate change we are looking for candidates willing to develop novel methodologies based on machine learning, deep learning, pattern recognition and artificial intelligence for information extraction, classification, target detection and change detection in long and dense timeseries of remote sensing images.
The candidate will be requested to deal with multi-/hyper-spectral images acquired by passive satellite sensors and/or Synthetic Aperture Radar (SAR) images acquired from active systems for Earth Observation. Among the others, data from ESA Copernicus (Sentinels), ASI PRISMA and COSMO-SkyMed will be considered. The goal is to design novel methods able to use temporal correlation to model landcover behaviors, changes and trends for a better understanding of phenomena over the past and the future for detecting trends and changes for modeling and understanding their impacts on climate and environment.
Besides the requirements established by the rules of the ICT school, preferential characteristics for candidates for this scholarship are:
•    master degree in Electrical Engineering, Communication Engineering, Computer Science, Mathematics or equivalents;
•    knowledge in pattern recognition, deep learning, image/signal processing, statistic/remote sensing, passive/active sensors.

Contact: Francesca Bovolo bovolo [at]