SOURCE-LINKED INTELLIGENCE

Position Matters: Feature Inversion Attacks in ViT Split Inference with Token Reduction and Shuffling

arXiv · AI, language, vision and robotics · article · Sep 1, 2026 · UTC

Vision Transformers (ViTs) are increasingly used in split-inference systems, where edge devices transmit intermediate token representations to a remote cloud. In this setting, token reduction lowers computation and communication costs, while token shuffling disrupts the spatial organization of the transmitted tokens, potentially limiting information leakage. However, their privacy benefits remain unclear against feature inversion attacks, which attempt to reconstruct the input from the transmitted embeddings. In this work, we show that, despite disrupting the spatial structure required by conv

Read original source ↗ Open in workspace

recordType

paper

region

Global