AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Deep Diffusion-Invariant Wasserstein Distributional Classification

Sung Woo Park, Dong Wook Shu, Junseok Kwon

NeurIPS 2020 · Label-efficient and robust learning › Recognition and long-tail · approach: Loss design and classifier theory · 0 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 DeepWDC instead of ResNetCIFAR10 classifier · CIFAR10 {θ, sc, sh, ϵ}accuracy: 73.6 -> 88.6Improved
use DeepWDC instead of ResNetCIFAR10 classifier · CIFAR10 {θ2, sc, sh, ϵ}accuracy: 73.4 -> 93.0Improved
use DeepWDC instead of DenseNetCIFAR10 classifier · CIFAR10-Caccuracy: 92.6 -> 95.8Improved
use DeepWDC instead of DenseNetCIFAR10 classifier · CIFAR10 {θ2, cc2}accuracy: 74.8 -> 87.7Improved
use DeepWDC instead of DGCNNModelNet10 classifier · ModelNet10 {T}accuracy: 97.1 -> 96.9No clear effect
use DeepWDC instead of DGCNNModelNet10 classifier · ModelNet10 {T, s, ϵ}accuracy: 83.6 -> 94.8Improved
use DeepWDC instead of DGCNNModelNet10 classifier · ModelNet10 {T, s, ϵ3, θ}accuracy: 35.4 -> 55.2Improved
use DeepWDC instead of PointNet++ModelNet10 classifier · ModelNet10 {T, s, ϵ3, θ}accuracy: 25.8 -> 55.2Improved
remove the diffusion invariance term, keeping only the intrinsic distance term (ablation)DeepWDC · CIFAR10 with perturbationsaccuracyHurt
vary covariance dimension d (16, 64, 128) and rank (8, 32); use d=128 rank-32 (ablation)DeepWDC · CIFAR10 with perturbationsaccuracyImproved

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