AI Paper Atlas

Research map / Optimization / Optimizers

Improving Gradient Flow with Unrolled Highway Expectation Maximization

Chonghyuk Song, Eunseok Kim, Inwook Shim

AAAI 2021 · Optimization › Optimizers · approach: Constrained and non-backprop training · 1 citations (OpenAlex)

Read the paper

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
Replace the EM M-step with a Newton-Raphson GEM update that adds a scaled skip connectionEM attention with ResNet-101 · PASCAL VOC val (SS, 256 channels)mIoU: 79.73 -> 80.93Improved
Replace the EM M-step with a Newton-Raphson GEM update that adds a scaled skip connectionEM attention with ResNet-101 · PASCAL VOC testmIoU: 87.7 -> 88.0Improved
Replace the EM M-step with a Newton-Raphson GEM update that adds a scaled skip connectionEM attention with ResNet-101 · PASCAL Context testmIoU: 53.1 -> 54.3Improved
Replace the EM M-step with a Newton-Raphson GEM update that adds a scaled skip connectionEM attention with ResNet-101 · COCO Stuff testmIoU: 39.9 -> 40.1Improved
Use intermediate step sizes rather than eta_train 1.0 (ablation)HEMNet · PASCAL VOCmIoU: 77.55 -> 78.22Improved
Increase T_train from 1 to 2 at eta_train 0.8 (ablation)HEMNet · PASCAL VOCmIoU: 77.26 -> 78.10Improved
Increase T_train beyond the performance peak (ablation)HEMNet · PASCAL VOCmIoU: 78.22 -> 77.83Hurt
Increase T_eval beyond T_train after training with eta_train 0.8 (ablation)HEMNet · PASCAL VOCmIoU: 78.10 -> 78.42Improved
Increase T_eval from 4 to 6 after training with eta_train 0.2 (ablation)HEMNet · PASCAL VOCmIoU: 77.96 -> 78.22No 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