AI Paper Atlas

Research map / Learning theory and algorithms / Clustering

Clustered Influence Functions

Miklós Máté Badó, Kristian Fenech

ICML 2026 · Learning theory and algorithms › Clustering · approach: Clustering approximation algorithms

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
Cache clustered IHVP solves and reuse across folds instead of per-query IFCiF, LeNet · CIFAR-10 K=50+100speedup: 32.1 h IF vs 2.1 h CiF, 15.3×Improved
Cache clustered IHVP solves instead of per-query IFCiF, AlexNet · CIFAR-10 K=50+100speedup: 63.3 h IF vs 5.4 h CiF, 11.7×Improved
Increase cache size from C=10 to C=50 (ablation)CiF, LeNet · CIFAR-10 K=50+100speedup: 15.3× -> 3.0×Hurt
Increase cache size from C=10 to C=50 (ablation)CiF, LeNet/AlexNet · CIFAR-10 K=50Pearson agreement: 0.943 -> 0.942; 0.994 -> 0.994No clear effect
Increase cache size from C=10 to C=50 (ablation)CiF, ResNet-18 · CIFAR-10 K=50Pearson agreement: 0.740 -> 0.935Improved
Increase cache size from C=128 to C=512 (ablation)CiF, DeiT-Tiny · ImageNet-1k K=5000rank correlation rho: 0.07 -> 0.19Improved
Increase cache size from C=128 to C=512 (ablation)CiF, DeiT-Tiny · ImageNet-1k K=5000speedup: 32.9× -> 9.3×Hurt
Add CiF cache on top of LoGra backendLoGra · CIFAR-10 ResNet-18 online queriesmean time per query: 8.68 ms -> 3.45 msImproved
Aggregate gradients by clusters for backdoor detection with fixed-location triggerCiF (C=80) · MNIST backdoor, fixed triggerpoison recallImproved
Aggregate gradients by clusters for backdoor detection with random-location triggerCiF (C=80) · MNIST backdoor, random triggerpoison recallHurt

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