Reserved topic scholarships - 2019 - 1st call

Department of Information Engineering and Computer Science

A1 - Open-ended Learning for Social Robots (1 grant)

The main goal of this EUREGIO funded project is teaching robots to perform a great variety of tasks, including collaborative tasks, and tasks not specifically foreseen by its designers.  Thus, the space of potentially-important aspects of perception and action is by necessity extremely large, since every aspect may become important at some point in time.  Conventional machine learning methods cannot be directly applied in such unconstrained circumstances, as the training demands increase with the sizes of the input and output spaces. As such, this research will be on novel machine learning paradigms including self-paced learning and  Multi-task Generative Adversarial Networks for object recognition and human behavior perception.
Contact: niculae.sebe [at] unitn.itenver.sangineto [at]

A2 - Grounded Conversational Agents (1 grant)

Conversational agents are gaining again quite some attention thanks also to the impressive results that have been obtained by Neural Network (NN) models. In particular, it has been shown that NNs are able to learn to carry out conversations about images. We are interested in zooming into the quality of the multimodal representations the  grounded conversational agents learn to compute through interaction. We conjecture that to reach good multimodal representation, the models should be trained on different tasks in parallel or incrementally.  
Contact: raffaella.bernardi [at]

A3 - Conversational Modeling and Systems (1 grant)

The PhD fellowship will support research in the most advanced conversational models of human-machine learning models that are effective and explainable to human experts. The research will be validated in the real-user scenarios as part of the ongoing research project at Signals and Interactive System Lab at the Department of Information Engineering and Computer Science
Contact: giuseppe.riccardi [at]

B1 - Cyber ranges as a multi-domain platform for security research and training (1 grant)

A Cyber range is a vitualization platform providing computers, networks  and systems on which various real-world cyber threat scenarios can be evaluated to provide a comprehensive, unbiased assessment of the security of Information and Industrial Control systems (IT and CTS). Cyber range must provide the capabilityof data collection, aggregation, correlation and automatic replay for the scenario owner or any "specialized user" to review attacks-defense processes on known targets ans future zero-day research. The student will work in the framework of the CyberSec4Europe Pilot Project (WP3, WP6, WP7) with other European partners to investigate key research and training idead so that they can be used in practice.
Contact: fabio.massacci [at]

B2 - Next-generation Ultra-wideband Localization and Communication for the Internet of Things (1 grant)

Research will focus on protocols and techniques to improve the data rate, energy efficiency, reliability and scalability of solutions based on ultra-wideband (UWB) radios. These radios can be used for communication but also for accurate distance estimation (ranging) and therefore localization. The proposed research will focus on both aspects, communication and localization, as well as their integration.
Contact: gianpietro.picco [at]

C1 - Learning human behaviour from streams of personal data (1 grant)

The goal of this thesis is to develop a platform for learning from personal data streama, and to use it in large scale real world experiments. The platform will have to be able to deploy state of the art machine learning algorithms in a robust manner. The data collected will be used to inform the machine learning algorithms which, in turn, will have to create user friendly services
Contact: Fausto.Giunchiglia [at]

C2 - Design and realization of a GDPR compliant data infrastructure (1 grant)

The topic of this PhD thesis will cover inter-disciplinary topics at the boundary between computer science, social sciences and privacy. The goal is to develop a general methodology for the design of person-centric data collection pilots and for the follow-up exploitation of these data in experiments. the methodology will have to be GDPR compliant and make data available according to the state-of-the-art best practices in data management
Contact: Fausto.Giunchiglia [at]

D2 - Development of methodologies and automatic techniques for the analysis of data acquired by planetary radars (1 grant)

The research activities related to this PhD position are related to planetary radar sounders that are instruments for the study of the subsurface of celestial bodies. These radars operate from satellite platforms and acquire data related to the subsurface. The activity will be focused on the definition, design, implementation and validation of:
1) radar signal processing algorithms;
2) data analysis techniques for supporting the automatic generation of products for the ground segment of planetary missions;
3) radar acquisition strategies for addressing the geophysics and geological challenges of a given mission.
Part of the research will be related to the activities in progress on the  Radar for Icy Moon Exploration (RIME) developed in the framework of the JUpiter ICy moon Explorer (JUICE) mission of the European Space Agency. RIME is developed under the leadership of the University of Trento in cooperation with Italian Industry and the Jet Propulsion Laboratory (JPL)  in USA with the funding of Italian Space Agency and NASA, respectively. RIME is a radar sounder defined to study the geology, the geophysics and the possible presence of water in the subsurface (up to 9 km) of the Jupiter icy moons, i.e., Ganymede, Europa and Callisto. For more information on RIME refer to
The PhD student will work at the Remote Sensing Laboratory of the University of Trento.
For more information on the activity of RSLab refer to: 
Contact: lorenzo.bruzzone [at]

Fondazione Bruno Kessler (FBK)

A4 - Deep Learning for Urban Environment (1 grant)

This PhD project has the ambition to explore the fusion of multiple modalities and sources of information and the design of novel cross-modal deep neural network architectures to study urban environments and social behaviors. In addition, the candidate will work on innovative Generative Adversarial Network models to generate realistic urban patterns and social behaviors that capture the great diversity of urban forms and lifestyles across the globe.
The ideal candidate will be strongly motivated in developing skills in machine learning with a special focus on deep learning, and in computer vision and multimodal approaches. The project will be supervised by Bruno Lepri (FBK) and Elisa Ricci (FBK/DISI).
Contact: lepri [at]

A5 - Deep Learning, constraints and network topologies (1 grant)

This PhD project will draw upon the recent extensive literature on human collective decision making and behavior and on network science as a source of inspiration for improving current deep learning architectures. For example, the student will explore if alternative network topologies (i.e. topologies encoding specific social constraints, sparser topologies, etc.) can improve deep neural network training leading to learning improvements at lower computational costs. The student will be supervised by Bruno Lepri (FBK) and co-supervised by Andrea Passerini (DISI).
Contact: lepri [at]

A6 - Incremental learning of abstract planning models via acting in a real environment (1 grant)

Autonomous agents, such as robots, chat-bots, self-driving cars, soft-bots etc., need to plan their actions in order to achieve their goals. For this reason, they should know the environment in which they operate and the effects of their actions on the environment. These information are usually encoded in the so-called “planning domain”, which, need to be “programmed off line” when the agent is programmed. However, the environment is dynamic and can have unpredicted changes; therefore, the agent should be able to adapt to unexpected situations. Furthermore, the effect of actions could be vary complex and unknown since the beginning; the agents should be able to learn action effects while acting. The objective of the Ph.D is to develop the necessary theory and the algorithms that allow an agent to incrementally learn a discrete, compact, and semantically rich representation of the planning domain in an environment in which it is supposed to interact. This representation is formulated in a form of a discrete planning domain.
Contact: serafini [at]
traverso [at]

A7 - Bayesian reasoning for statistical relational learning (1 grant)

Current approaches in statistical relational learning are based on undirected graphical models such as Markov Logic Networks. State of the art algorithms for statistical inference cover the Maximum Likelihood (ML) and Maximum a Posteriori (MAP) tasks, but not so much attention has been devoted to Bayesian Inference. Due to the high complexity of the models that can be generated, statistical inference is approximated using sampling methods.  Recently, we proposed a study about Bayesian Inference in hybrid graphical models (i.e., models composed of discrete and continuous random variables); the advantage of Bayesian inference is that, it’s a truly statistical inference and it is very robust to overfitting training data. We design a variational method to solve the “exact inference”. However, to perform Bayesian inference, combinatorial problems on the discrete variables must be solved in a more efficient way, and this is still an open problem. The objective of this thesis, is to extend such proposals and to make scalable. 
Contact: serafini [at]

A8 - End-to-End Automatic Speech Recognition (1 grant)

Sequence to sequence modeling for automatic speech recognition (ASR), through deep learning approaches, is now state of the art. Several architectures have been proposed, and related software tools are available,  for copying with this task.
Nevertheless, ASR performance still presents serious drawbacks in the presence of high noise and reverberation levels in the input signal, as well as  when voices of multiple speakers  overlap.
Possible methods to overcome these limitations is to provide ASR systems with several, possibly independent, information sources, e.g.  by means of microphone arrays or using video information.
The objective of the PHD will be to advance the present state of the art in this field especially focusing on recent application scenarios, such as dinner party and distant reverberant  ASR.
To do this possible research topics could investigate methods for: selecting the best channel, fuse information in the sequence encoded by the DNN, improve  the attention model employed by the decoder, ... .
Contact: falavi [at]

A9 - Neural speech-translation (1 grant)

Recent advances in deep learning are giving the possibility to address traditional NLP tasks in a new and completely different manner. One of these tasks is spoken language translation (SLT), which combines the main challenges of its two parent research areas: automatic speech recognition (ASR) and machine translation (MT). For years, SLT has been addressed by cascading an ASR and an MT system. Recent trends rely on using a single neural network to directly translate the input audio signal in one language into a text in a different language without intermediate transcription steps. This approach has several advantages over traditional cascaded solutions: it reduces the engineering required to train separate modules, it avoids their cumulative errors and it allows to directly use speech prosodic cues to improve the translation. Although the number of publications at top conferences and journals is increasing, SLT technology is still far behind its parent tasks in terms of final performance. More research is hence required to make SLT architectures closer to production deployment. This PHD will focus on developing cutting-edge solutions able to advance the state of the art in SLT. Possible research topics include the design of methods to improve the audio signal representation in the encoder, the use of weakly supervised learning techniques, the efficient adaptation of existing models to new speakers or domains.
Contact: turchi [at]

A10 - Fast and high-precision 3D inspection and monitoring of non-collaborative surfaces (1 grant)

3D optical metrology inspection and monitoring solutions of industrial parts, composed of non-collaborative surfaces, relies on very expensive technologies based on active sensors, such as laser scanning arms, structured light systems, confocal white light, etc. A viable and efficient solution based on pure (monocular / stereo) imaging and vision methods is not yet reliable and applicable in the production line due to the various problems of image-based methods in case of reflective, translucent and shining objects. The investigation should analyze and evaluate existing methods and find innovative image-based solutions to solve actual challenges in surface inspection of non-collaborative surfaces, such as glasses, mirrors, cars bodywork, reflective plastics, etc. New algorithms and methods based e.g. on multispectral imaging, Visual SLAM, semi-global matching, etc. will be considered and assembled.
Contact: remondino [at]

B3 - Programmable 5G systems (1 grant)

Although 4G seems to have just arrived, the transition towards 5G mobile networks has already started. 5G aims at enabling a connected world where multiple verticals (e.g., automotive, industry, and health), each characterized by different performance targets in terms of bitrate, latency, and reliability, can coexist on the same infrastructure. This scenario calls for advanced isolation and resource management mechanisms. Several technologies have recently emerged to address these challenges. Radio Access Network (RAN) slicing, a non-native concept of 4G, enables dynamic partitioning of dedicated RAN resources into isolated virtual networks. Conversely, Mobile Edge Computing (MEC), allows off-loading computationally intensive tasks to computing nodes located in the RAN and thus very close the mobile terminals. In this PhD the student will study novel design for a 5G network for autonomous and connected vehicles combining MEC and network slicing. More specifically, the PhD student will apply slicing and MEC techniques in the context of real-time management of swarms of vehicles (e.g. drones or cars).
Contact: rriggio [at]

B4 - Application-aware fog computing (1 grant)

Fog computing extends cloud computing technology to the edge of the infrastructure to support dynamic computation for IoT applications. Reduced latency and location awareness in objects’ data access is attained by displacing workloads from the central cloud to edge devices, within the so-called cloud-to-edge continuum. By doing so, fog computing enables low-latency and privacy-preserving processing of data close to where it is actually produced and consumed. In addition to that, it overcomes communication bottlenecks and reduces costs, representing a key step towards the pervasive uptake of IoT/edge-based services. This PhD project has the ambition of investigating dynamic resource orchestration methods and algorithms to smartly deploy applications in fog computing environments, with the aim of catering to the needs of applications while optimising the utilisation of the infrastructure. The work will cover both theoretical and practical aspects.
Contact: dsiracusa [at]

B5 - Stretchable antennas (1 grant)

Antennas are key components in communications that radiate electromagnetic waves which can be used to carry information. With respect the traditional technologies, a stretchable antenna is deformable, it means that the shape and resistance change thus changing the radiation properties in general. The main objective of this research is to investigate stretchable antennas designs and the related fabrication technologies. Finally, performance results in terms of attainable strain (stretchability), conductivity, antenna efficiency, and cost will be studied and discussed for application in wearable electronics and sensing.
Contact: lorenzel [at] fbk.eumulloni [at]

C3 - Default in contextualized knowledge representation (1 grant)

Representation of context has been one of the main approaches in the knowledge representation and reasoning area for the management of large knowledge bases: here, modelling and reasoning on knowledge is relativized with respect to the contexts (situations, set of hypothesis) in which it is supposed to hold. Introducing forms of non-monotonic and default reasoning in logic-based contextual frameworks is a challenging research issue which, in particular, plays a role in the propagation and inheritance (with overriding) of knowledge across hierarchies of contexts. The objective of this PhD will be to investigate the possibilities for extending the theory and implementation of the current methods for representing default information in contextual frameworks and explore their applications.
Contact: bozzato [at]

C4 - Fusion of remote sensing and citizen science information for geospatial city sensing (1 grant)

Within the Flagship project “City Sensing”, SCC (Smart City and Communities Impact Line) is looking for a PhD candidate who is interested in designing methodologies for the fusion and the exploitation of multi-source data and modeling techniques to better predict and understand cities’ characteristics and outcomes, and to provide solutions to different challenges of citizens’ daily lives.
The candidate is expected to work within the area of City Science with the ambition of using satellite, georeferenced and crowdsourced (e.g., OpenStreetMap) BigData for solving city challenges related to security, mobility, etc. Methodologies will be designed in the fields of artificial intelligence, deep learning, data fusion.
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, Geographic Information Science (GIScience), or equivalents;
• interdisciplinary knowledge in pattern recognition, image/signal processing, statistic/remote sensing.
Contact: bovolo [at] fbk.eunapolitano [at]

C5 - Formal and genetic methods for model-based testing of parameterized systems (1 grant)

Many moders software systems are highly parameterized in order to take into account various operational conditions. Testing such problem is extremely difficult, given the astronomic number of combinations corresponding to all the parameter assignments. The challenge is to maximize the coverage and the confidence provided by the generated test cases without incurring in the limitations deriving from the parameterization space. The thesis will investigate the definition of novel testing techniques starting from the integration genetic  algorithms for software testing and test case generation approaches based on formal methods.
Contact: cimatti [at]

C6 - Safety analysis for space and avionic systems and software (1 grant)

Space and avionic systems are reaching an unprecedented degree of complexity. The process of safety analysis attempts to characterize the likelyhood of faults and failures, and to assess the effectiveness of the adopted mitigation measures. Unfortunately, traditional techniques are becoming ineffective, unable to deal with large scale systems.
This thesis will investigate novel methods for safety analysis, based on the adotpion of formal models of system and software (nominal and faulty) behaviours. Particularly interesting are the analysis of iming aspects in the propagation of multiple faults to failures and errors, the ability to explain the causality of propagation, and the definition of techniques for on-the-fly fault detection, isolation and recovery policies.
Contact: cimatti [at]

C7 - Condition monitoring for predictive maintenance by integrating machine learning and background knowledge (1 grant)

The current trend of automation and data exchange in manufacturing technologies allows to gather huge amounts of information. One of the challenges is to assess the condition of an equipment and foresee a failure before it actually occurs, so that fine-tuned, predictive maintenance can be applied. Although machine learning techniques can be used to tackle this problem, a pure data-driven approach is bound to re-discovering background information that is already available, such as operation modes and procedures, failure modes, and maintenance records. The thesis will investigate the integration of symbolic background knowledge within modern machine learning approaches, in order reduce the training time and improve the accuracy of predictions.
Contact: cimatti [at]

D3 - Design of methods for automatic analysis of sub-surface radargrams (1 grant)

In the context of the development of the Radar for Icy Moons Exploration (RIME) payload on board of European Space Agency (ESA) JUpiter ICy moons Explorer (JUICE), the Remote Sensing for Digital Earth Unit (RSDE) at Fondazione Bruno Kessler (FBK) is developing research activities in the design of methods for automatic analysis of radargrams. The goal is to model, extract and understand sub-surface features and targets in radar sounder data (radargrams). Satellite-borne radar sounders have been successfully employed to analyze the sub-surface of the Moon and Mars and airborne instruments have been operated for Earth observation. RIME will explore Jupiter Icy Moons, and an instrument is under investigation for Venus exploration. In this context, the PhD research activity will be focused on the design and development of radar signal processing methods, automatic algorithms for sub-surface information extraction, radargrams classification and target detection in radargrams by means of machine learning and pattern recognition methodologies.
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, image/signal processing, statistic/remote sensing/radar.
Contact: bovolo [at]

D4 - Satellite Image Time Series (SITS) analysis (1 grant)

Remote Sensing for Digital Earth Unit (RSDE) at Fondazione Bruno Kessler (FBK) is seeking for a PhD student willing to work on the analysis of Satellite Image Time Series (SITS). The candidate will work on the design of machine-learning-, deep-learning-, artificial-intelligence-based methodologies for automatic information extraction and retrieval in SITS. Multitemporal information extraction requires to design methods able to deal with Bigdata acquired by different kinds of sensors (ESA Sentinels and older ones). The activity will be developed by considering applications like environmental monitoring, climate change, city sensing, etc.
Besides the general 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, image/signal processing, machine learning/deep learning/remote sensing.
Contact: bovolo [at]

OSRAM - Department of Information Engineering and Computer Science

A11 - Domain adaptation for people detection, re-identification and pose estimation (1 grant)

Computer vision and machine learning solutions for people detection and re-identifications have been proposed in the past but the main limitation is that they depend on specific data which the models were trained on. To address this, the project addresses research on domain adaptation in order to investigate the models, learning and data strategies which enable generalization. We will consider algorithms for people detection, re-identification and pose estimation, which we will generalize to novel environments, e.g. different people and illumination, different cameras and optics, and different setups.  The research is  relevant to resilient-safety city, as it may support the identification/tracking of wanted individuals or accessing limited-access areas.
The project is jointly funded by UNITN and OSRAM. As such we will consider also the successful innovation transfer to a number of industrial applications directly relevant to OSRAM, including, but not limited to, office and retail.
Contact: niculae.sebe [at]
e.ricci [at] unitn.itfabio.galasso [at]

University of Trento - Department of Information Engineering and Computer Science

D1 - Development of methodologies and automatic techniques based on artificial intelligence and machine learning for the analysis of satellite remote sensing images (1 grant)

The activity that will be developed is related to the definition, the design, the development and the validation of advanced artificial intelligence and machine learning methods for the analysis of big data from space. The activities will address different topics related to the main methodological problems to be solved for the analysis of the huge archives of remote sensing data that are acquired by Earth Observation satellites. The main topics include:
1) deep learning architectures for the automatic classification (semantic segmentation) of satellite images;
2) data analysis techniques for the automatic extraction of semantic  information from data;
3) data fusion techniques for the integration of multisensor and multisource data. Part of the developed methods will be applied to the analysis of remote sensing data related to problems of climate change (for more information on the related project refer to 
The PhD student will work at the Remote Sensing Laboratory of the University of Trento.
For more information on the activity of RSLab refer to: 
Contact: lorenzo.bruzzone [at]