AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

LinGCN: Structural Linearized Graph Convolutional Network for Homomorphically Encrypted Inference

Hongwu Peng, Ran Ran, Yukui Luo, Jiahui Zhao, Shaoyi Huang, Kiran Thorat, Tong Geng, Chenghong Wang, Xiaolin Xu, Wujie Wen, Caiwen Ding

NeurIPS 2023 · Architectures and efficiency › RNNs and normalization · approach: Efficient network compression and training · 10 citations (OpenAlex)

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
use LinGCN instead of CryptoGCN at 6 nonlinear layersSTGCN-3-128 · NTU-XViewtest accuracy (%): 74.25 -> 77.55Improved
use LinGCN instead of CryptoGCN at 6 nonlinear layersSTGCN-3-128 · NTU-XViewlatency (s): 4273.89 -> 1856.95Improved
use LinGCN instead of CryptoGCN at 6 nonlinear layersSTGCN-3-256 · NTU-XViewtest accuracy (%): 75.31 -> 80.29Improved
use LinGCN instead of CryptoGCN at 6 nonlinear layersSTGCN-3-256 · NTU-XViewlatency (s): 10580.41 -> 4632.05Improved
reduce the effective nonlinear layers from 6 to 2LinGCN STGCN-3-128 · NTU-XViewlatency (s): 1856.95 -> 741.55Improved
reduce the effective nonlinear layers from 6 to 2LinGCN STGCN-3-128 · NTU-XViewtest accuracy (%): 77.55 -> 75.16Hurt
reduce the effective nonlinear layers from 6 to 4LinGCN STGCN-6-256 · NTU-XViewtest accuracy (%): 84.52 -> 85.78Improved
reduce nonlinear layers from 6 to 2LinGCN · Flickrlatency (s): 4290.93 -> 2740.94Improved
reduce nonlinear layers from 6 to 2LinGCN · Flickraccuracy (val/test, %): 0.5281/0.5275 -> 0.5247/0.5266Hurt
perform nonlinear reduction before polynomial replacement (ablation)LinGCN STGCN-3-256 · NTU-XViewaccuracy (%): more than 2%Improved
use structural rather than layer-wise linearization (ablation)LinGCN STGCN-3-256 · NTU-XView, 2 to 5 effective nonlinear layersaccuracy (%): 1.5%Improved
set distillation KL weight η to 0.2 (ablation)LinGCN STGCN-3-256, 6 nonlinear layers · NTU-XViewaccuracy (%)Improved

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