HaMMon and HaMMon advance
HaMMon
The project aims to develop tools and methodologies to be used in different industrial contexts for the quantification of the impacts of extreme natural events on the Italian territory. In order, also, to support activities focused on monitoring, forecasting and assessment of environmental risks and protection from potentially critical phenomena. The reference areas are the identification of factors contributing to the risk, forecasts on the occurrence of extreme events in the medium term and the quantification of impacts after catastrophic events. The activities will involve intensive use of scientific visualization and artificial intelligence technologies, especially for assessing and extracting meaningful information on risk-exposed assets.
The aims of the project are:
– Improvement of risk mapping related to extreme natural events for Italy
– Enhancement of monitoring activities of the Italian territory, aimed at quantifying
the impacts of extreme events
– Improvement of knowledge about assets exposed to catastrophic risks
– Development of high-resolution medium range forecasts of extreme events in Italy
– Development of a technological infrastructure to process and distribute data
HaMMon-Advance: Integrated AI for Enhanced Seasonal Forecasting and Environmental Risk Management
This project expands the HaMMon initiative by improving its system for seasonal forecasting and hazard assessment of extreme events. It will refine the built environment classification algorithms initially developed in WP3 and WP4 of HaMMon. The project is intended to improve efficiency in the processing and analysis of environmental data, with the goal of achieving more accurate predictions of extreme weather events and a clearer understanding of the vulnerability of the built environment to climate-induced hazards.
La collaborazione di HaMMon e HaMMon-Advance
Spoke Proponente: Spoke 3
Aziende: UnipolSai, Sogei, IFAB (International Foundation Big Data and Artificial Intelligence for Human Development)
Soggetti Pubblici: INAF, INFN, IREA,ENEA, FBK, CMCC, Politecnico di Bari, Università di: Torino, La Sapienza, L’Aquila, Bari, Salento e Trento
Interoperable Data Lake (IDL)
The Project aims at creating a Data Lake service, supporting a seamless access to space and ground-based observations and simulated data. The project addresses the design and commissioning of an interoperable, distributed data archive, relying on state-of-the-art open technologies, supporting both science and industry
The service will specifically address the challenges related to the big data scenario, in terms of both data management, storage, access, identification and of access to computing resources necessary to process the data.
The definition of a data model to describe and access the data will be designed according to FAIR and IVOA standards.
New techniques of block-chain and web-based stacks like Object Storage, will be used, tuned and linked together to optimize big data storage and efficiency of data retrieval, exploiting state-of-the-art cloud-based technologies.
A high-level functional architecture of the system (space + ground) will be deployed, ensuring the suitable level of secure access (SSA), focusing in particular on the processing of satellite data.
Effective algorithms for data processing will be selected to create a mock-up simulator capable of generating a synthetic data set based on the identified use cases and operational scenarios, testing the algorithmic chain, and evaluating the computational load.
La collaborazione di IDL
Spoke Proponente: Spoke 3
Aziende: Leonardo S.p.A., Thales Alenia Space Italia S.p.A.
Soggetti Pubblici: INAF, INFN
Fraud Detection
In finance, fraud detection is an extremely important form of anomaly detection. Some examples are identifying fraudulent credit card transactions and financial documents. These tasks are typically performed using machine/deep learning techniques.
Researchers have proposed several quantum versions of these approaches, which may provide a substantial speed-up and increase the quality of the outcomes.
At the same time, there is interest in approaching the same problem via classic methods, in order to provide a comparison between Quantum vs Non Quantum solutions.
La collaborazione di FRAUD DETECTION
Spoke Proponente: Spoke 10
Azienda: Banca Intesa Sanpaolo
Soggetti Pubblici: INAF, Università Roma Tor Vergata, Università di Padova, Università di Firenze, Università La Sapienza, Politecnico di Miliano
Time Series in the banking sector
The aim of the project consists in applying ML techniques to solve problems in the banking sector, usually described as time series (TS).
The peculiarity of this TS is that they are quite sparse because they are based on events (e.g. transactions) and the target estimation could be a classification of an event (fraud, not fraud) or a probability for a certain event to happen (e.g. default, not default). Moreover, the target could be highly unbalanced as proposed in the following banking case: data quality, corporate credit risk, churn analysis. The dataset of Data Quality will be the most important, the other two, if needed and available, could be used to test the algorithm in a different domain
La collaborazione di Time Series in the banking sector
Spoke Proponente: Spoke 3
Azienda: Banca Intesa Sanpaolo
Soggetti Pubblici: INAF, Università Roma Tor Vergata
Serial Code Porting on HPC & Quantum Computing
The goal of the project is to rewrite Machine Learning (ML) algorithms, available in their Open Source versions, in a suitable language for HPC computation as well as in modules compatible with architectures based on quantum computation.
The project objective is system development, based on ML algorithms in order to prevent cyber attacks (ref. KeepCalm Project).
HPC or quantum computing-based environments would allow real-time response to cyber attacks, increasing reactive capacity, a fundamental factor in cybersecurity.
Project Short Description
The proposed PoC is focused on the following areas:
● Use of algorithms in the HPC and Quantum Computing fields.
● Creation of a methodology used for a systemic rewriting of ML algorithms relevant
for business activities.
● Clustering activity (within HPC architectures) of algorithms used for the R&D
world.
La collaborazione di Serial Code Porting on HPC & Quantum Computing
Spoke Proponente: Spoke 3
Azienda: Sogei
Soggetti Pubblici: INAF, Università Roma Tor Vergata
Harnessing the power of Artificial Intelligence for predictive maintenance of industrial plants
Demonstrate the usability of modern AI-based techniques in the realization of systems for predictive maintenance and for the modeling of the interdependencies between systems in complex industrial apparatuses.
Techniques in use in the academic domain (for example those on the detection of anomalous signals, those parsing log entries from IT systems, those dealing with graphs) will be tested on appropriate data samples coming from ENI productive sites; these data can range from sensors reading in industrial facilities, single apparatus data, complex relationship graphs between multiple apparatuses.
La collaborazione di Harnessing the power of Artificial Intelligence for predictive maintenance of industrial plants
Spoke Proponente: Spoke 2
Azienda: Sogei
Soggetti Pubblici: ENI, Università di Bologna, INAF e INFN
Anomalies in time series (ATS)
Anomaly detection identifies rare items, events, or patterns deviating significantly from the majority of data. Techniques for time series anomaly detection include clustering-based techniques, density-based techniques, prediction-based techniques, and more. Banking time series aim to detect suspicious events and data processing issues. Astrophysics anomaly detection targets rare events and instrumental issues. The project aims at developing, in both domains, a robust, cross-applicable solution for time series anomaly detection.
La collaborazione di ATS
Spoke Proponente: Spoke 3
Azienda: Banca Intesa Sanpaolo
Soggetti Pubblici: INAF e Università di Trieste