AI Paper Atlas

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

Scalable Infomin Learning

Yanzhi Chen, Weihao Sun, Yingzhen Li, Adrian Weller

NeurIPS 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Contrastive representation theory · 1 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 adversarial MI estimators with sliced (Slice) estimatorinfomin representation learning · US Census Demographicfairness ρ*(Z,T): 0.16 \pm 0.10 -> 0.07 \pm 0.02Improved
use Slice instead of Neural Rényiinfomin representation learning · UCI Adultfairness ρ*(Z,T): 0.17 \pm 0.08 -> 0.08 \pm 0.02Improved
use Slice instead of dCorr/Pearson non-parametric penaltiesinfomin representation learning · US Census Demographicfairness ρ*(Z,T): 0.47 \pm 0.08 -> 0.07 \pm 0.02Improved
apply Slice infomin terminfomin representation learning · Dspriteρ*(Z,T): 0.91 \pm 0.03 -> 0.08 \pm 0.02Improved
apply Slice infomin terminfomin representation learning · Dspritereconstruction MSE: 0.37 \pm 0.01 -> 0.50 \pm 0.01Hurt
apply Slice infomin terminfomin representation learning · CMU-PIEρ*(Z,T): 0.76 \pm 0.04 -> 0.07 \pm 0.01Improved
apply Slice infomin terminfomin representation learning · CMU-PIEreconstruction MSE: 1.81 \pm 0.04 -> 2.15 \pm 0.07Hurt
use Slice instead of adversarial/Neural TCinfomin representation learning · MNIST->MNIST-Mtarget accuracy: 46.3 \pm 0.03 -> 98.5 \pm 0.45Improved
use Slice instead of Neural TCinfomin representation learning · MNIST->MNIST-Mρ*(Zc,Zd): 0.64 \pm 0.04 -> 0.06 \pm 0.01Improved
use Slice instead of Neural TCinfomin representation learning · CIFAR10->STL10target accuracy: 80.8 \pm 0.08 -> 82.3 \pm 0.03Improved
apply Slice infomin terminfomin representation learning · CIFAR10->STL10source accuracy: 93.0 \pm 0.03 -> 92.5 \pm 0.03Hurt
apply Slice infomin terminfomin representation learning · US Census Demographicutility ρ*(Z,Y): 0.95 \pm 0.00 -> 0.95 \pm 0.01No clear effect

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