AI Paper Atlas

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

AIM: Manifold-based Data Filtering for Representation Finetuning

Qing Li, Qibin Zheng, Yi Liu, Xingchun Diao

AAAI 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Multilingual language representation pretraining · 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 AIM manifold-based filtering instead of LESSReFT · Commonsense avg (LLaMA-3.2-3B)average accuracy: 51.21 -> 54.62Improved
use AIM instead of LossSelReFT · Commonsense avg (LLaMA-2-7B)average accuracy: 45.00 -> 47.42Improved
use AIM instead of Random on LLaMA-3-8BReFT · Commonsense avg (LLaMA-3-8B)average accuracy: 64.12 -> 67.33Improved
use LossSel or LESS instead of random selection on LLaMA-3-8B commonsenseReFT · Commonsense avg (LLaMA-3-8B)average accuracy: 64.12 -> 61.71 (LossSel); 61.46 (LESS)Hurt
use AIM instead of LossSelReFT · Arithmetic avg (LLaMA-3-8B)average accuracy: 74.87 -> 76.67Improved
use AIM instead of RandomReFT · Arithmetic avg (LLaMA-3.2-3B)average accuracy: 60.10 -> 62.00No clear effect
use AIM-C classifier approximation instead of full AIM (ablation)AIM · Arithmetic avg (LLaMA-2-7B)average accuracy: 57.21 -> 57.62No clear effect
use AIM-C instead of full AIM (ablation)AIM · Commonsense avg (LLaMA-3-8B)average accuracy: 67.33 -> 62.08Hurt
weighted vs uniform segment aggregation, single segments (ablation)AIM score · Commonsense selection precisionprecisionMixed
set explained variance threshold rho to 0.90 (ablation)AIM · Commonsense selection precisionprecision: peak at rho = 0.90Improved
ReFT vs LoRA under label noiseReFT · Commonsense reasoning, LLaMA-3 8Baverage accuracy under noiseHurt

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