GAN You See Me? Enhanced Data Reconstruction Attacks against Split Inference
Dec 10, 2023
Speakers
Mengda Yang
Řečník · 0 sledujících
Hongxin Hu
Řečník · 0 sledujících
Xiaoyang Xu
Řečník · 0 sledujících
About
Split Inference (SI) is an emerging deep learning paradigm that addresses computational constraints on edge devices and preserves data privacy through collaborative edge-cloud approaches. However, SI is vulnerable to Data Reconstruction Attacks (DRA), which aim to reconstruct users' private prediction instances. Existing attack methods suffer from various limitations. Optimization-based DRAs do not leverage public data effectively, while Learning-based DRAs depend heavily on auxiliary data quantity and distribution similarity. Consequently, these approaches yield unsatisfactory attack results and are sensitive to defense mechanisms. To overcome these challenges, we propose a GAN-based LAtent Space Search attack (GLASS) that harnesses abundant prior knowledge from public data using advanced StyleGAN technologies. Additionally, we introduce GLASS++ to enhance reconstruction stability. Our approach represents the first GAN-based DRA against SI, and extensive evaluation across different split points and adversary setups demonstrates its state-of-the-art performance. Moreover, we thoroughly examine seven defense mechanisms, highlighting our method's capability to reveal private information even in the presence of these defenses.Split Inference (SI) is an emerging deep learning paradigm that addresses computational constraints on edge devices and preserves data privacy through collaborative edge-cloud approaches. However, SI is vulnerable to Data Reconstruction Attacks (DRA), which aim to reconstruct users' private prediction instances. Existing attack methods suffer from various limitations. Optimization-based DRAs do not leverage public data effectively, while Learning-based DRAs depend heavily on auxiliary data quant…
Organizer
NeurIPS 2023
Účet · 646 sledujících
Like the format? Trust SlidesLive to capture your next event!
Professional recording and live streaming, delivered globally.
Sharing
Recommended Videos
Presentations on similar topic, category or speaker
Differentiable sorting for censored time-to-event data.
Zhlédnout později
Pro uložení prezentace do věčného trezoru hlasovalo 0 diváků, což je 0.0 %
Resource-rational prediction in real and artificial neural networks
Zhlédnout později
Pro uložení prezentace do věčného trezoru hlasovalo 0 diváků, což je 0.0 %
Convolutional Neural Operators for robust and accurate learning of PDEs
Zhlédnout později
Pro uložení prezentace do věčného trezoru hlasovalo 0 diváků, což je 0.0 %
Near-Optimal Bounds for Learning Gaussian Halfspaces with Random Classification Noise
Zhlédnout později
Pro uložení prezentace do věčného trezoru hlasovalo 0 diváků, což je 0.0 %
Ambient Diffusion: Learning Clean Distributions from Corrupted Data
Zhlédnout později
Pro uložení prezentace do věčného trezoru hlasovalo 0 diváků, což je 0.0 %
Slow Science
Zhlédnout později
Milind Tambe, …
Pro uložení prezentace do věčného trezoru hlasovalo 0 diváků, což je 0.0 %