AI Paper Atlas

Research map / Federated learning and privacy / Personalized and heterogeneous FL

Preservation of the Global Knowledge by Not-True Distillation in Federated Learning

Gihun Lee, Minchan Jeong, Yongjin Shin, Sangmin Bae, Se-Young Yun

NeurIPS 2022 · Federated learning and privacy › Personalized and heterogeneous FL · approach: Non-IID federated learning · 42 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
add not-true distillation loss to local trainingFedAvg · CIFAR-10 sharding s=2Accuracy@1 (%): 40.14 -> 52.61Improved
add not-true distillation lossFedAvg · CIFAR-10 LDA a=0.05Accuracy@1 (%): 28.24 -> 40.17Improved
add not-true distillation lossFedAvg · CINIC-10 sharding s=2Accuracy@1 (%): 39.64 -> 48.07Improved
add not-true distillation lossFedAvg · MNIST sharding s=2Accuracy@1 (%): 78.63 -> 84.44Improved
add not-true distillation lossFedAvg · CIFAR-100 LDA a=0.1Accuracy@1 (%): 30.69 -> 32.37Improved
use FedNTD instead of SCAFFOLDSCAFFOLD · CIFAR-10 sharding s=10Accuracy@1 (%): 68.97 -> 68.56Hurt
use FedNTD instead of SCAFFOLDSCAFFOLD · CIFAR-10 LDA a=0.1Accuracy@1 (%): 23.12 -> 54.42Improved
increase local epochs from 1 to 20FedNTD · CIFAR-10 sharding s=2Accuracy@1 (%): 35.77 -> 60.61Improved
increase local epochs to 20FedNTD vs SCAFFOLD · CIFAR-10 sharding s=2Accuracy@1 (%): 62.80 -> 60.61Hurt
lower client sampling ratio to 0.05FedNTD · CIFAR-10 sharding s=2Accuracy@1 (%): 33.06 -> 46.99Improved
reduce forgetting measure via not-true distillationFedNTD · CIFAR-10 sharding s=2forgetting F: 0.59 -> 0.43Improved
add not-true distillation on out-local classesFedNTD · CIFAR-10 NIID s=2weight alignment (Linear_1): 0.635 -> 0.703Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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