AI Paper Atlas

Research map / Federated learning and privacy / Privacy attacks and secure learning

SecMoE: Communication-Efficient Secure MoE Inference via Select-Then-Compute

Bowen Shen, Yuyue Chen, Peng Yang, Bin Zhang, Xi Zhang, Zoe L. Jiang

AAAI 2026 · Federated learning and privacy › Privacy attacks and secure learning · approach: Encryption-based private learning · 1 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
select expert obliviously then compute only one expert (Select-Then-Compute) instead of evaluating all expertssecure MoE inference vs BumbleBee · MoE-Small LAN, 128 expertsruntime (min): 9.76 -> 2.52Improved
Select-Then-Compute MoE layer vs Ironsecure MoE inference · MoE-Small, 128 experts (communication)communication (GB): 24.17 -> 0.82Improved
Select-Then-Compute vs BumbleBee on WANsecure MoE inference · MoE-Small WAN, 128 expertsruntime (min): 13.88 -> 3.68Improved
increase experts from 32 to 128 in SecMoE (ablation)SecMoE · MoE-Small LANruntime (min): 2.04 -> 2.52No clear effect
replace BumbleBee GeLU with Select-Then-Compute piecewise quadratic GeLU (ablation)SecMoE · Switch-Base 128 expertsGeLU communication: 81% less communication, 7.1x speedupImproved
replace BumbleBee GeLU with Select-Then-Compute GeLU (ablation)SecMoE · MoE-Small 128 expertsGeLU communication / LAN runtime: 44% less communication, 11% less LAN runtimeImproved
use secure SecMoE protocol instead of plaintext inferenceMoE-Small · CoLA/QNLI/RTE validationMCC / accuracy: QNLI 90.3 -> 90.2; RTE 69.9 -> 70.0; CoLA 41.0 -> 41.0No clear effect

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