Multi-fidelity, adaptive technique in design-exploration space, ANOVA-analysis

Docente

Ludovico Biagi (mail)

Area di ricerca

Geoinformatica

Keyword (max 3 separate da virgola)

Prediction, samplets

Tecnologie da utilizzare

python, C++, binding, I/O on dataset

Descrizione (max 500 caratteri)

Goal This project aims to develop additional features, e.g. multi-fidelity, adaptive technique in design-exploration space, ANOVA-analysis. The results will be tested on industrial testcase of WindAssistant Trasoceanic Cargo, Energy System and on geoinformatics data (e.g. GNSS dataset).
Project Plan :
• Read and understand a reference paper on samplets, kernel-based methods, multifidelity (provided)
• Getting familiar with Bayesian Optimization, Anova, MultiFidelity algorithms
• Design, train and implement the studied algorithms
• Performance assessment on an industrial dataset as well as on geoinformatics dataset
• Cowriting a report for scientific dissemination in ECCOMAS conference and for peer-review paper
Scroll to Top