Conference proceeding
FAME: Adaptive Functional Attention with Expert Routing for Function-on-Function Regression
39th Conference on Neural Information Processing Systems, NeurIPS 2025, Vol.38, pp.68447-68476
2025
DOI: 10.52202/085713-2057
Abstract
Functional data play a pivotal role across science and engineering, yet their infinite-dimensional nature makes representation learning challenging. Conventional statistical models depend on pre-chosen basis expansions or kernels, limiting the flexibility of data-driven discovery, while many deep-learning pipelines treat functions as fixed-grid vectors, ignoring inherent continuity. In this paper, we introduce Functional Attention with a Mixture-of-Experts (FAME), an end-to-end, fully data-driven framework for function-on-function regression. FAME forms continuous attention by coupling a bidirectional neural controlled differential equation with MoE-driven vector fields to capture intra-functional continuity, and further fuses change to inter-functional dependencies via multi-head cross attention. Extensive experiments on synthetic and real-world functional regression benchmarks show that FAME achieves state-of-the-art accuracy and strong robustness to arbitrarily sampled discrete observations of functions.
Details
- Title: Subtitle
- FAME: Adaptive Functional Attention with Expert Routing for Function-on-Function Regression
- Creators
- Yifei Gao - Tsinghua UniversityYong Chen - Department of Industrial and Systems Engineering, University of Iowa, Iowa City, IA, United StatesChen Zhang - Tsinghua University
- Resource Type
- Conference proceeding
- Publication Details
- 39th Conference on Neural Information Processing Systems, NeurIPS 2025, Vol.38, pp.68447-68476
- DOI
- 10.52202/085713-2057
- ISSN
- 1049-5258
- Grant note
- 20243080039 / Tsinghua University (501100004147) 72271138 / National Natural Science Foundation of China (http://data.elsevier.com/vocabulary/SciValFunders/501100001809) 20243080039 / Tsinghua-NUS Joint Funding 72271138 / National Natural Science Foundation of China (501100001809)
- Language
- English
- Date published
- 2025
- Academic Unit
- Industrial and Systems Engineering
- Record Identifier
- 9985222684302771
Metrics
1 Record Views