AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Efficient Activation Function Optimization through Surrogate Modeling

Garrett Bingham, Risto Miikkulainen

NeurIPS 2023 · Architectures and efficiency › RNNs and normalization · approach: Theory of depth and width · 2 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
replace ReLU with AQuaSurF-discovered HardSigmoid(HardSigmoid(x))*ELU(x)All-CNN-C · CIFAR-100accuracy: 0.6897 -> 0.6990Improved
replace ReLU with discovered Swish(-2x)ResNet-56 · CIFAR-100accuracy: 0.7348 -> 0.7469Improved
replace ReLU with discovered -x*sigma(x)*HardSigmoid(x)MobileViTv2-0.5 · ImageNetaccuracy: 0.6139 -> 0.6396Improved
transfer discovered -x*sigma(x)*HardSigmoid(x) to ResNet-50 instead of ReLUResNet-50 · ImageNettop-1 accuracy: 0.7660 -> 0.7776Improved
transfer discovered x*erfc(ELU(x)) to ResNet-50 instead of ReLUResNet-50 · ImageNettop-1 accuracy: 0.7660 -> 0.7635Hurt
replace ReLU with discovered erfc(Softplus(x))^2 (sigmoidal)CoAtNet · Imagenetteaccuracy: 0.8772 -> 0.8907Improved
use sigmoid instead of ReLU as activationAll-CNN-C · CIFAR-100accuracy: 0.6897 -> 0.0100Hurt
use Swish instead of ReLUMobileViTv2-0.5 · ImageNetaccuracy: 0.6139 -> 0.6383Improved
extend search with ELiSH baseline instead of ReLUCoAtNet · Imagenetteaccuracy: 0.8772 -> 0.1000Hurt
use FIM eigenvalues plus function outputs instead of either feature alone for surrogate (ablation)KNR/RFR/SVR search · Act-Bench datasetsbest validation accuracy foundImproved
use UMAP-feature regression instead of random search (ablation)KNR/RFR/SVR search · Act-Bench datasetsbest validation accuracy foundImproved
use KNR instead of RFR or SVR as search algorithm (ablation)AQuaSurF · Act-Bench datasetsbest validation accuracy foundImproved

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