AI Paper Atlas

Research map / Optimization / Optimizers

Proximal Mean-Field for Neural Network Quantization

Thalaiyasingam Ajanthan, Puneet K. Dokania, Richard Hartley, Philip H. S. Torr

ICCV 2019 · Optimization › Optimizers · approach: Constrained and non-backprop training · 15 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 PMF instead of BinaryConnectVGG-16 binary network · CIFAR-100top-1 test accuracy: 43.70 -> 61.52Improved
use PMF instead of BinaryConnectVGG-16 binary network · CIFAR-10top-1 test accuracy: 86.40 -> 90.51Improved
use PMF instead of BinaryConnectResNet-18 binary network · TinyImageNettop-1 test accuracy: 49.33 -> 51.00Improved
use PMF instead of ProxQuantResNet-18 binary network · CIFAR-100top-1 test accuracy: 68.35 -> 71.85Improved
use PMF instead of BinaryConnectLeNet-5 · MNISTtop-1 test accuracy: 99.30 -> 99.44No clear effect
add softmax projection (noisy, annealed) instead of sparsemax projection (ablation)PMF · CIFAR-100 VGG-16top-1 test accuracy: 57.83 -> 61.52Improved
add softmax projection instead of sparsemax projection (ablation)PMF · CIFAR-10 ResNet-18top-1 test accuracy: 92.60 -> 92.73Improved
add softmax projection with annealed beta instead of hardmax (proximal ICM) (ablation)PMF · CIFAR-100 VGG-16top-1 test accuracy: 45.65 -> 61.52Improved
add softmax projection instead of hardmax (ablation)PMF · TinyImageNet ResNet-18top-1 test accuracy: 49.66 -> 51.00Improved
use proximal ICM (hardmax, u-space) instead of BinaryConnectbinary network · CIFAR-10 / CIFAR-100 / TinyImageNettop-1 test accuracy: 86.40 -> 88.96Improved
use PMF on CIFAR-100 ResNet-18 instead of VGG-16 (gap to float reference)PMF · CIFAR-100top-1 test accuracy: 61.52 -> 71.85Improved

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