Advanced Computing and Electromagnetics
The Advanced Computing and Electromagnetics Research Area is split in two Research Unit dedicated to tree main topics:
- applied research activities in Advance Computing with the study and design of distributed computing architecture based on public and private platforms like AWS, Openstack. Low power architetture in high performance perspective, orchestration of heterogenous architecture in computing continuum, resources and application management in distributed environment;
- applied research and activities for the design and prototyping antennas. Electromagnetic compatibility, experience in computational electromagnetism;
- applied research in distributed database technologies and data mining algorithms for Big data and extreme scale analysis
Main focus and assets on Advance Computing topic are:
• Low power high performance architecture
• Hardware/software integration for microdatacenter solution in low power computing e low power communication perspective
• Machine learning: CNN
• Application acceleration, deep learning (CNN, BNN) on reconfigurable architecture (FPGA)
• Application porting on OpenCL framework through hardware accelerator FPGA (Intel/Altera) and GPU
• Application analysis, design and development for machine learning, HPC-embedded context through many-core architecture
• Offloading solution from edge to datacenters.
Main focus and assets on Computational Elettromagnetism topic are:
• Design and prototypting antenna
• Electromagnetic simulation and modellization
• Metamaterials
• Fast solver for CEM application
• Low Power communication strategies
• Tools for antenna measurments and diagnostic.
Main focus and assets on distributed database technologies are:
• Distributed Data model design for heterogeneous data sources in Big data application context
• Fine-grained real time analysis and prediction of buildings thermal energy consumption using data from smart meters
• Ontological modeling of cultural heritage data from heterogeneous sources for NLP
• Characterization of user interests from social networks data (e.g., about cultural heritage)
• Big Data distributed processing engines: Apache Spark, MongoDB clusters with MapReduce
• Relational and NoSQL databases, RDF Triple Stores
• Machine Learning frameworks: Spark MLLib, scikit-learn, R, Rapid Miner
• Supervised learning algorithms: regression and classification (ANN, SVM, Decision Tree, Random Forest, LR-SGD); unsupervised learning algorithms: clustering, association rules mining.
