AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

Cross-modal Scalable Hyperbolic Hierarchical Clustering

Teng Long, Nanne van Noord

ICCV 2023 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Self-supervision in supervised pipelines · 9 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
split hyperbolic hierarchical clustering into Sinkhorn pre-clustering plus hyperbolic post-clustering (ablation)HypHC-style clustering · Kinetics-Sound multi-modalNMI: 20.6 -> 36.9Improved
add hyperbolic hierarchical post-clustering on top of Sinkhorn pre-clustering (ablation)SeLaVi · Kinetics-Sound multi-modalNMI: 36.4 -> 36.9Improved
use hyperbolic space instead of Euclidean for the hierarchy (ablation)sHHC · Kinetics-Sound multi-modalNMI: 36.3 -> 36.9Improved
move SeLaVi to hyperbolic space without hierarchy (ablation)SeLaVi · Kinetics-Sound multi-modalNMI: 36.4 -> 36.1No clear effect
use fewer clusters (34 instead of 100) (ablation)sHHC · Kinetics-SoundNMI: 36.9 -> 37.4Improved
use sHHC hierarchy instead of HypHC or UFithierarchical clustering · VGGSound hierarchyDasgupta cost / runtime: 7.72 -> 6.62; ~4h -> ~2min (vs UFit)Improved
use sHHC hierarchical pseudo-labels for pretrainingL3 backbone · KineticsSound multi-modal recognitionaccuracy: 70.95 -> 74.10Improved
use half-precision or mixed precision traininghyperbolic clustering · numerical stabilityperformanceHurt

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