Permutation Equivariance of Transformers and Its Applications

• Published in: arXiv.org
• Main Authors: Xu, Hengyuan, Liyao Xiang, Ye, Hangyu, Yao, Dixi, Chu, Pengzhi, Li, Baochun
• Format: Paper
• Language: English
• Published: Ithaca Cornell University Library, arXiv.org 31.03.2024
• Subjects: Coders; Equivalence; Learning; Neural networks; Permutations; Privacy; Training; Transformers

Revolutionizing the field of deep learning, Transformer-based models have achieved remarkable performance in many tasks. Recent research has recognized these models are robust to shuffling but are limited to inter-token permutation in the forward propagation. In this work, we propose our definition of permutation equivariance, a broader concept covering both inter- and intra- token permutation in the forward and backward propagation of neural networks. We rigorously proved that such permutation equivariance property can be satisfied on most vanilla Transformer-based models with almost no adaptation. We examine the property over a range of state-of-the-art models including ViT, Bert, GPT, and others, with experimental validations. Further, as a proof-of-concept, we explore how real-world applications including privacy-enhancing split learning, and model authorization, could exploit the permutation equivariance property, which implicates wider, intriguing application scenarios.