Research
Universal Approximation of Operators with Transformers and Neural Integral Operators
arXiv:2409.00841v3 Announce Type: replace Abstract: We study the universal approximation properties of transformers and neural integral operators for operators in Banach spaces. In particular, we show
arXiv:2409.00841v3 Announce Type: replace Abstract: We study the universal approximation properties of transformers and neural integral operators for operators in Banach spaces. In particular, we show that the transformer architecture is a universal approximator of integral operators between Holder spaces. Moreover, we show that a generalized version of neural integral operators, based on the Gavurin integral, are universal approximators of arbitrary operators between Banach spaces. Lastly, we show that a modified version of transformer, which uses Leray-Schauder mappings, is a universal approximator of operators between arbitrary Banach spaces.
Related
- Quantitative Approximation Rates for Group Equivariant Learning
- MENO: MeanFlow-Enhanced Neural Operators for Dynamical Systems
- Kernel Neural Operators (KNOs) for Scalable, Memory-efficient, Geometrically-flexible Operator Learning
- Time-Frequency Analysis for Neural Networks
- Certified and accurate computation of function space norms of deep neural networks
Source: arXiv cs.LG | 2026-04-28