Embedding finite-volume HLLC flux conservation into CNN training cuts airfoil drag prediction error substantially versus pure pixel MAE, especially in low-data regimes.
18th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference , year=
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
CoFINN: Conservation Flux Informed Neural Networks for Physics Problems Governed by Conservation Laws
Embedding finite-volume HLLC flux conservation into CNN training cuts airfoil drag prediction error substantially versus pure pixel MAE, especially in low-data regimes.