The macro-area is composed of two main research areas that are transversal with respect to the methodological approach, attentive to the foundational aspects of research topics deriving from the computational sciences for life. The first area focuses on the study of theoretical and foundational aspects of computation, modeling, and simulation of complex systems and the development of innovative algorithmic methodologies: areas motivated by application contexts such as biology and physics that require the need to define computational foundations. In this context, there are two main specific objectives. The first focuses on the design and experimentation of exact, approximate, evolutionary algorithmic methodologies or machine learning aimed at processing large amounts of data in various application contexts, mainly related to the analysis and integration of biological data. The second objective focuses on the study of computational power and properties of computational models inspired by physical and biological laws, such as membrane systems (or P systems) and reaction systems, with the definition of associated complexity classes and comparison with classical models such as Turing machines and RAM machines, and on the extension and study of specific compositional properties of models, such as Petri nets. The application areas in this context include biology and various aspects of the simulation of complex systems under uncertain conditions, safety and control. The second area of research is in the computational sciences for life and addresses various algorithmic and computational aspects of genomics, transcriptomics and phylogenetics in Bioinformatics, simulation and modelling of processes in computational biology and specifically systems biology. The activities in this area are broad and complementary. The main lines of research active in the field of fundamentals are:
- study of the algebraic structure and logical properties of space of local states of competing systems,
- calculation models for the description of real phenomena and systems (in particular cellular automata),
- Introduction into machine learning of aspects of uncertainty, such as soft clustering, generalized decision trees, inference on possibilistic databases,
- study of the computational power and computational efficiency of membrane systems (P systems) and reaction systems,
- algorithms and succinct data structures for indexing and querying large collections of strongly similar texts,
- general-purpose computing with GPU.
In life sciences and bioinformatics, the main lines of active research are:
- event prediction algorithms for splicing events from NGS data, NGS data assembly and algorithms for the reconstruction and comparison of phylogenetic trees with cancer genomics applications,
- algorithms and models for inferencing the evolution of tumor mutations from large quays of single-cell or bulk sequencing data,
- deterministic and stochastic simulation algorithms for biological systems,
- mechanistic and constraint-based models of biological systems,
- machine learning and deep learning for the classification of tumor subtypes.
Recently is active in the development of algorithms in Fundamentals a new direction of research on Information Security mainly focused on the development and implementation of algorithms and cryptographic protocols.
The area deals with various research topics related to the development of software systems, with particular attention to design, quality control, and maintenance and evolution of software systems. The research activity on software design mainly concerns with the definition, development and testing of new architectural models and appropriate methodologies and tools to address emerging application domains. The research activity on software quality mainly concerns with the definition of methods for the validation and verification of software systems, the definition of testing techniques, the dynamic and static analysis of programs and the realization of self-healing systems. Finally, the research activity on software systems maintenance mainly concerns with the definition, development, and validation of reverse engineering techniques and tools for software comprehension, maintenance and technical debt management. The research activity carried out in this area is wide and varied, the main lines of research in progress are:
- the definition and experimentation of architectural solutions for multimodal systems,
- the definition and experimentation of self-healing and runtime enforcement techniques for Mobile and Cloud environments,
- the definition of techniques for software monitoring, recognition and prediction of failures, and automatic repair of faults, particularly in Cloud and Mobile environments,
- automatic generation of unit and system test cases using symbolic, search-based and machine learning techniques,
- the development of techniques to identify and quantify the technical debt level of a software project and avoid the degradation of the software architecture,
- the study of machine learning techniques for the recognition of code and design anomalies and the definition of the corresponding refactoring techniques to solve them,
- the development of tools to support the modernization of legacy systems architecture towards microservice architecture.
The research activities carried out in this macro-area relate to Data Science, and are aimed at developing models and techniques to support the processes of management and analysis of various types of data. In particular, multiple aspects related to the data life cycle are considered, including acquisition, transformation, organization, different types of analysis, knowledge extraction, and interaction with users.
Two main lines of research can be distinguished: the first includes Information Retrieval, Text Mining, and social media analysis, while the second includes the intersection of contiguous and complementary disciplines such as Human-Machine Interaction, Information Systems, Distributed Systems, and Data Semantics. In the first research area, the main interests relate to the definition of models, techniques, and systems aimed at guaranteeing personalized access to information on the Web, the analysis and conceptual representation of texts, the extraction of information from texts (text mining), the analysis of the evolution of information, and the predictive analysis of user-generated content in social media. In the second research line, problems concerning three macro-themes are addressed, often jointly: a) human-machine interaction, with particular attention to human interaction with artifacts, graphic interface design, and data visualization; b) semantic interoperability between systems for data, knowledge and services management, with particular attention to the implementation of open, adaptive and distributed applications in the Cloud through service architectures, semantic integration, and quality analysis; c) value in information systems, with particular attention to the value of information and services. The most active lines of research at this time, include:
Information Retrieval Area:
- Definition of models and systems for personalized search and recommendation of information in various application contexts.
- Definition of models for the semantic representation of texts and for the analysis of the evolution of knowledge over time on social media.
- Definition of user models based on natural language analysis.
- Analysis of information generated by users in social media
Area Human-Machine Interaction and Distributed and Semantic Information Systems:
- Design, Analysis, Development and Usability Assessment of visual interactive systems for the exploration and analysis of large amounts of data.
- Semantic enrichment techniques of structured data to support Data Science applications.
- API composition techniques based on semantic descriptions.
- Profiling models, quality and value of data and services.
The activity in this area focuses on foundational and industrial research topics covering 1) diverse aspects of the analysis and management of mutimedia and sensorial data of several kinds, 2) robotics, and 3) real-time intelligent sensing. The research activities on topic 1) focuses on the development of algorithms and techniques for analysing and managing digital signals, with special interest on multimedia signals (audio, images, and video), fisiological signals (EEG (electroencefalogram), ECG (electrocardiogram), skin galvanic responses, blood pressure, temperature), and psychofisical signals (eye tracker).
Moreover, it designs and develops novel methods for Computer Vision, Pattern Recognition, Machine Learning, Artificial Intelligence, and Multimedia, applied to diverse types of data with a specialization on managing images, video, and data from multimodal sensors. The research activity on robotics focuses on systems for world-perception by autonomous robots, in particular mapping the working environment, localization, and scene understanding and tracking. The research on real-time systems is developed over conceptual and computational instruments for understanding and controlling complex systems which evolve with time, passing through a series of macroscopic states, each of which is characterised by a specific set of rules which depend on data collected at real-time.
The main active research topics are:
- processing, analysis, and classification of multimedia signals in the realm of affective signal processing and analysis of social networks;
- integration of fisiological multimodal data for classification of cognitive and emotive states, or identifying potential pathologies (for example, prosopagnosia; the incapacity of recognising faces) or age-related diseases;
- Internet of Things and Autonomous Driving;
- cross-media execution in robotics; for example, localization with cameras in LIDAR-constructed maps;
- Wearable Expery System Development, integration between knowledge-based systems and wearable technologies for constructing recommender systems in time-dependent domains;
- use of machine learning techniques, in particular case-based reasoning, for the analysis and support of decisions in domains with heterogenerous data.
This research area studies research topics traditionally linked to Artificial Intelligence along methods, techniques, models, and applications for decision support. In particular, inference models based on techniques for knowledge representation and management, and statistical inferences. Learning models and algorithms based on structured, semi-structured, and unstructured data are designed and experimented, along with computational methods for optimization problems for data analytics. Also distributed approaches (multi-agent systems) for modelling and simulating complex systems, natural or artificial, are designed.
Within the large spectrum of research challenges associated to this area, the main research lines are:
- design, development, and application of knowledge-based systems for decision support;
- definition of learning and inference techniques for heterogeneous data generated in relational contexts, integrating content and structure analysis (social network analysis, text analytics);
- definition of global stochastic optimization techniques and reinforcement learning for optmizing the hyperparameters in machine learning and automated configuration;
- development of models and algorithms for automated reasoning and optimization for data analytics (text analytics and text mining, time series analysis, clustering);
- development and application of algorithm for structural and parametric learning for continuous time Bayesian networks with applications to genetic regulatory networks;
- development and experimenting with complex natural or artifical systems based on agent-based models (for example, vehicle or pedestrian traffic).