FP=XINT: Representing Neural Networks via Low-Bit Series Basis Functions
Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (33) · 28168-28175
DOI: 10.1609/aaai.v40i33.40043 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Boyang Zhang und 5 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref).
Abstract (Originalsprache)
Deep neural networks are often over-parameterized, resulting in prohibitive storage and computational costs. A fundamental question is whether a complex network can be re-expressed in terms of a compact set of basis functions without sacrificing accuracy. Motivated by this perspective, we aim to approximate a dense model by decomposing it into a small number of lightweight components that capture the essential functional structure of the network. To this end, we propose a series expansion framework that rewrites a neural network as a linear combination of low-bit basis models. Within the post-training quantization setting, the full-precision model is expanded hierarchically at the tensor, layer, and model levels into a structured set of basis functions. We theoretically prove that this expansion converges exponentially to the original model. Furthermore, we design AbelianAdd and AbelianMul operations between isomorphic basis models, endowing the expansion with an Abelian group structure that naturally supports commutative and parallel computation. Experimental results across diverse architectures show that our series expansion method leverages a set of ultra-low-bit basis functions, not only preserving full-precision performance without the need for calibration data or fine-tuning, but also featuring a parallel-friendly design that enables efficient and scalable deployment.
Zitieren
APA
Zhang, B., Cheng, D., Zhang, Y., Tian, J., Li, J., & Liu, F. (2026). FP=XINT: Representing Neural Networks via Low-Bit Series Basis Functions. Proceedings of the AAAI Conference on Artificial Intelligence, 40(33), 28168-28175. https://doi.org/10.1609/aaai.v40i33.40043
Vancouver
Zhang B, Cheng D, Zhang Y, Tian J, Li J, Liu F. FP=XINT: Representing Neural Networks via Low-Bit Series Basis Functions. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(33):28168-28175. doi:10.1609/aaai.v40i33.40043
BibTeX
@article{Zhang2026,
author = {Zhang, Boyang and Cheng, Daning and Zhang, Yunquan and Tian, Jiake and Li, Jing and Liu, Fangming},
title = {FP=XINT: Representing Neural Networks via Low-Bit Series Basis Functions},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2026},
volume = {40},
number = {33},
pages = {28168--28175},
doi = {10.1609/aaai.v40i33.40043}
}
EUVIMED