11 papers
Capacity Overflow: A Blind Spot for Backdoor Attacks in Vision MoE
Xiaocheng Zou, Tiancheng Zheng, Xiaolin Xu +1
Mixture-of-Experts (MoE) has become a prevalent paradigm for scaling Vision Transformers efficiently. To ensure computational scalability and prevent expert overload, Vision MoE ar…
Groundhog Bit-Flip Attack: Seeding Infinite Generation Loops in Mixture-of-Experts LLMs through Bit Flips
Huakang Lin, Tiancheng Zheng, Mingxuan Sun +4
Mixture-of-Experts (MoE) architectures enable scalable and efficient large language models (LLMs) by selectively activating expert sub-networks through a routing mechanism. However…
SLAC: Access-Driven CPU-to-GPU Side-channel Attacks via System-Level Cache on Apple Silicon
Tianhong Xu, Saion K. Roy, Ruyi Ding +2
Modern heterogeneous System-on-Chip designs integrate CPU cores and a GPU that share a last-level cache (LLC) or system-level cache (SLC). This sharing exposes a new cross-domain a…
RouteScan: A Non-Intrusive Approach to Auditing MoE LLMs Safety via Expert Routing Telemetry
Bo Lv, Zhiheng Xu, KeDong Xiu +4
As Mixture-of-Experts (MoE) architectures are increasingly adopted for scaling Large Language Models (LLMs), safety auditing becomes necessary to verify whether these models produc…
MetaSeal: Defending Against Image Attribution Forgery Through Content-Dependent Cryptographic Watermarks
Tong Zhou, Ruyi Ding, Gaowen Liu +5
The rapid growth of digital and AI-generated images has amplified the need for secure and verifiable methods of image attribution. While digital watermarking offers more robust pro…
MoEcho: Exploiting Side-Channel Attacks to Compromise User Privacy in Mixture-of-Experts LLMs
Ruyi Ding, Tianhong Xu, Xinyi Shen +2
The transformer architecture has become a cornerstone of modern AI, fueling remarkable progress across applications in natural language processing, computer vision, and multimodal…
MACPruning: Dynamic Operation Pruning to Mitigate Side-Channel DNN Model Extraction
Ruyi Ding, Cheng Gongye, Davis Ranney +2
As deep learning gains popularity, edge IoT devices have seen proliferating deployment of pre-trained Deep Neural Network (DNN) models. These DNNs represent valuable intellectual p…
Graph in the Vault: Protecting Edge GNN Inference with Trusted Execution Environment
Ruyi Ding, Tianhong Xu, Aidong Adam Ding +1
Wide deployment of machine learning models on edge devices has rendered the model intellectual property (IP) and data privacy vulnerable. We propose GNNVault, the first secure Grap…
Probe-Me-Not: Protecting Pre-trained Encoders from Malicious Probing
Ruyi Ding, Tong Zhou, Lili Su +3
Adapting pre-trained deep learning models to customized tasks has become a popular choice for developers to cope with limited computational resources and data volume. More specific…
Non-transferable Pruning
Ruyi Ding, Lili Su, Aidong Adam Ding +1
Pretrained Deep Neural Networks (DNNs), developed from extensive datasets to integrate multifaceted knowledge, are increasingly recognized as valuable intellectual property (IP). T…
GraphCroc: Cross-Correlation Autoencoder for Graph Structural Reconstruction
Shijin Duan, Ruyi Ding, Jiaxing He +3
Graph-structured data is integral to many applications, prompting the development of various graph representation methods. Graph autoencoders (GAEs), in particular, reconstruct gra…