AI Paper Atlas

Research map / Architectures and efficiency / CNN design

Memory-Optimal Direct Convolutions for Maximizing Classification Accuracy in Embedded Applications

Albert Gural, Boris Murmann

ICML 2019 · Architectures and efficiency › CNN design · approach: Compact CNN architectures · 17 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 herringbone instead of naive convolution (ablation)CNN conv layers · Activation memorybytes: 760.0 B -> 434.5 BImproved
use replace method instead of naive (ablation)CNN conv layers · Activation memorybytes: 760.0 B -> 467.5 BImproved
use herringbone instead of replace (ablation)CNN conv layers · Activation memorybytes: 467.5 B -> 434.5 BImproved
use single transpose instead of herringbone (ablation)CNN conv layers · Activation memorybytes: 434.5 B -> 437.5 BHurt
use single transpose instead of herringbone (ablation)CNN conv layers · Inference timeseconds: 819s -> 457sImproved
use naive or replace instead of herringbone (ablation)CNN conv layers · Inference timeseconds: 819s -> 396sImproved
peak stack usage with single transpose (ablation)CNN on Arduino · SRAM budgetspare bytes: 2B spareHurt
halve activation memory for given budget (naive vs herringbone) (ablation)CNN · Validation error vs memoryvalidation error: 2x errorImproved
replace to herringbone method (ablation)CNN · Validation error vs memoryrelative error: ≈10%Improved
decrease activation memory 10x below 1KB (ablation)CNN · Validation errorvalidation error: 10xHurt
2KB memory-optimal CNN instead of 2KB BonsaiBonsai · MNISTaccuracy (%): 94.38% -> 99.15%Improved
2KB CNN instead of 2KB ProtoNNProtoNN · MNISTaccuracy (%): 93.25% -> 99.15%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