Reproduce “Bayesian Assessments of Aeroengine Performance with Transfer Learning” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Pranay Seshadri, Andrew Duncan, George Thorne, Geoffrey Parks, Raul Vazquez Diaz, Mark Girolami
A Bayesian Gaussian Process framework was developed for aeroengine performance assessment, integrating spatial field approximation, robust area averaging, and a rigorous decomposition of uncertainty. The approach also includes a transfer learning methodology to improve predictions in sparsely instrumented contexts by leveraging data from more heavily instrumented scenarios, yielding more precise spatial estimates and area averages.
Key insights
- The proposed 'Bayesian area average' consistently outperforms traditional sector area-averaging, showing less scatter and faster convergence towards the true area average.
- Uncertainty can be rigorously decomposed into components attributable to sensor imprecision and spatial sampling, providing actionable guidance for optimizing instrumentation strategy.
- Transfer learning effectively leverages data from more heavily instrumented contexts or adjacent planes to reduce predictive uncertainty in sparsely measured regions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.