AI Paper Atlas

Research map / Learning theory and algorithms / Prediction and decision losses

Simplex Clustering via sBeta With Applications to Online Adjustment of Black-Box Predictions

Florent Chiaroni, Malik Boudiaf, Amar Mitiche, Ismail Ben Ayed

TPAMI 2024 · Learning theory and algorithms › Prediction and decision losses · approach: Provable post-hoc explanations

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
add sBeta scaling (delta=0.15) to Beta-based clustering (ablation)k-Betas · VISDA-Caccuracy: 19.8 -> 56.0Improved
add sBeta scaling (delta=0.15) to Beta-based clustering (ablation)k-Betas · SVHN->MNISTaccuracy: 51.5 -> 76.6Improved
add sBeta scaling on synthetic Dirichlet data (ablation)k-Betas · SimuNMI: 81.1 -> 79.2Hurt
use vertex init instead of k-means++ (ablation)k-sBetas · SVHN->MNISTaccuracy: 69.8 -> 76.6Improved
use vertex init instead of k-means++ (ablation)k-sBetas · VISDA-Caccuracy: 47.2 -> 56.0Improved
use vertex init instead of k-means++ (ablation)k-means · SVHN->MNISTaccuracy: 66.6 -> 68.9Improved
use vertex init instead of k-means++ (ablation)k-medoids · SVHN->MNISTaccuracy: 51.9 -> 71.3Improved
cluster simplex softmax instead of argmaxk-sBetas · VISDA-Caccuracy: 53.1 -> 56.0Improved
replace Euclidean k-means with k-sBetas on simplexk-means · VISDA-Caccuracy: 47.9 -> 56.0Improved
cluster simplex predictions instead of feature maps with k-meansk-means · VISDA-Caccuracy: 22.7 -> 47.9Improved
downsample prediction map by factor 8 before clustering (ablation)k-sBetas · GTA5 road->Cityscapes roadmIoU: 65.7 -> 65.7No clear effect
use method of moments instead of MLE-1000 for parameter estimation (ablation)k-sBetas · SVHN->MNISTrunning time (s): 64.32 -> 0.48Improved

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