AI Paper Atlas

Research map / Architectures and efficiency / Neural architecture search

Binarized Neural Architecture Search for Efficient Object Recognition

Hanlin Chen, Li’an Zhuo, Baochang Zhang, Xiawu Zheng, Jianzhuang Liu, Rongrong Ji, David Doermann, Guodong Guo

IJCV 2020 · Architectures and efficiency › Neural architecture search · approach: Efficient NAS search strategies · 22 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 performance-based search instead of PC-DARTS for binarized (PCNN) architectureBNAS · CIFAR-10 W1A32test accuracy (%): 95.12 -> 96.06Improved
use performance-based search instead of PC-DARTS for 1-bit architectureBNAS · CIFAR-10 W1A1test accuracy (%): 90.06 -> 93.28Improved
add UCB exploration term to 1-bit BNAS (ablation)BNAS (1-bit) · CIFAR-10 W1A1test error (%): 8.3 -> 6.72Improved
use performance-based operation pruning instead of PC-DARTS searchBNAS · CIFAR-10 W1A32search cost (GPU days): 0.18 -> 0.09375Improved
use performance-based operation pruning for full-precision searchBNAS (full-precision) · CIFAR-10test error (%): 2.78 -> 2.84No clear effect
increase initial channels for XNOR-based BNAS (44 channels)BNAS (XNOR) · CIFAR-10 W1A32test error (%): 5.71 -> 4.88Improved
increase initial channels for PCNN-based BNAS (48 channels)BNAS · CIFAR-10 W1A32test error (%): 3.94 -> 3.47Improved
increase warm-up epochs L from 3 to 5 (ablation)BNAS (PCNN) · CIFAR-10accuracy (%): 95.8 -> 96.06Improved
increase warm-up epochs L from 5 to 11 (ablation)BNAS (PCNN) · CIFAR-10accuracy (%): 96.06 -> 96.03; search cost 0.09375 -> 0.1687No clear effect
search binarized architecture with BNAS instead of using hand-designed ResNet-18 (PCNN)BNAS · ImageNet W1A32top-1 accuracy (%): 63.5 -> 71.3Improved
search 1-bit architecture with BNAS instead of ResNet-18 (BONN)BNAS (1-bit) · ImageNet W1A1top-1 accuracy (%): 59.3 -> 64.3Improved
binarize weights of searched network versus full-precision PC-DARTSBNAS · ImageNettop-1 accuracy (%): 75.8 -> 71.3Hurt

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