AI Paper Atlas

Research map / Optimization / Optimizers

Gradient Inversion Attacks Beyond SGD

Guangnian Wan, Gongfan Fang, Xinyin Ma, Xinchao Wang

ICML 2026 · Optimization › Optimizers · approach: Adam convergence and adaptive learning rates

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add analytical label recovery from optimizer updates (ablation)update-matching attack · ImageNet (Adam)PSNR (dB): 15.3358 -> 19.6252Improved
add the proposed update-matching objective (ablation)label-recovery attack · ImageNet (Adam)PSNR (dB): 11.4221 -> 19.6252Improved
add analytical label recovery (ablation)update-matching attack · ImageNet (Adam)LPIPS: 0.7175 -> 0.4614Improved
add update-matching objective (ablation)label-recovery attack · ImageNet (Adam)LPIPS: 0.8413 -> 0.4614Improved
use the proposed attack instead of HFGI when only Adam updates are visibleHFGI · ImageNet (Adam updates)PSNR (dB): 13.3472 -> 20.5985Improved
use the proposed attack instead of HFGI when only Adam updates are visibleHFGI · PACS (Adam updates)PSNR (dB): 12.4712 -> 19.0078Improved
use the update-matching attack instead of HFGI with raw SGD gradientsHFGI · ImageNet (SGD)PSNR (dB): 20.7997 -> 23.9195Improved
use the update-matching attack instead of HFGI with raw SGD gradientsHFGI · PACS (SGD)PSNR (dB): 19.2953 -> 24.8643Improved
use the proposed attack instead of IG under RMSPropIG · ImageNet (RMSProp)PSNR (dB): 11.8515 -> 20.4626Improved
use the proposed attack instead of IG under AdaGradIG · ImageNet (AdaGrad)PSNR (dB): 9.6361 -> 18.7233Improved
use the proposed attack instead of IG under SGD with momentumIG · ImageNet (Momentum)PSNR (dB): 10.6764 -> 17.0135Improved
pair update-matching with existing label inferenceupdate-matching objective · ImageNet (Adam)reconstruction qualityHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy