KNOs combine deep compositions of kernel integral operators with neural networks to define expressive kernels, delivering universal approximation for operator learning with geometric flexibility and roughly 10x fewer parameters than prior neural operators.
W ENDLAND , Scattered Data Approximation, Cambridge University Press, 2005
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Kernel Neural Operators (KNOs) for Scalable, Memory-efficient, Geometrically-flexible Operator Learning
KNOs combine deep compositions of kernel integral operators with neural networks to define expressive kernels, delivering universal approximation for operator learning with geometric flexibility and roughly 10x fewer parameters than prior neural operators.