Permutation Equivariance of Transformers and its Applications

作者：Hengyuan Xu, Liyao Xiang, Hangyu Ye, Dixi Yao, Pengzhi Chu, Baochun Li · 年份：2024 · DOI：10.1109/cvpr52733.2024.00572 · 被引用次数：13 · 研究领域：Computability, Logic, AI Algorithms、Quantum chaos and dynamical systems、Random Matrices and Applications

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 per-mutation 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. The code is available at https://github.com/Doby-Xu/ST