arXiv Machine Learning June 5, 2026 By Aarav Bedi (University of California, Berkeley)

Auditing Demonstration Curation Metrics: Action-Only Scorers Fail on the Structural Defects That Degrade Imitation Policies

Read the original on arXiv Machine Learning →

arXiv:2606. 05588v1 Announce Type: cross Abstract: Imitation-learning policies inherit the quality of the demonstrations they are trained on, and a growing set of curation metrics promise to score and filter low-quality demonstrations automatically.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.

arXiv Machine Learning

Jun 16

Phase-Localized Curation Does Not Help: A Negative Result on Per-Phase Metric Selection for Demonstration Filtering

arXiv:2606. 15064v1 Announce Type: new Abstract: Manipulation demonstrations have temporal phase structure, and a natural hypothesis is that demonstration-curation metrics should be applied within phases rather than globally.

By Aarav Bedi

robotics

arXiv Machine Learning

Aug 24

Rethinking Demonstration Unlearning in Imitation Learning for Robotics

arXiv:2608.20784v1 Announce Type: cross Abstract: Imitation learning for robotics depends on human demonstrations, some of which people may later ask to remove. Retraining without them is the natural...

By Jiazhuo Li, Yu Zhang, Yiming Fei, Kangkang Dong, Xiaojun Zhu, Houde Liu, Jinze Tao

robotics

arXiv Machine Learning

Jun 10

What Demonstration Curation Metrics Do to Your Policy

arXiv:2606. 10229v1 Announce Type: cross Abstract: We study whether demonstration-curation metrics that detect defective training episodes also improve the downstream behavior-cloning policy that trains on the curated data.

By Aarav Bedi

benchmarks

arXiv Machine Learning

Aug 11

Auditing Instruction-Trajectory Mismatches in Multimodal Robot Demonstrations

arXiv:2608. 07895v1 Announce Type: cross Abstract: Robot demonstration datasets used to train vision-language-action policies can contain a subtle but harmful failure mode: trajectories that are behaviorally correct but paired with the wrong language instruction.

By Simon Holk, Ryosuke Takanami, Tatsuya Matsushima, Yusuke Iwasawa, Yutaka Matsuo, Yueh-Hua Wu, Kei Ota

robotics multimodal benchmarks

arXiv AI

Aug 6

RESample: A Robust Data Augmentation Framework via Exploratory Sampling for Robotic Manipulation

arXiv:2510. 17640v4 Announce Type: replace-cross Abstract: Vision-Language-Action (VLA) models have shown strong manipulation capability when trained with large-scale imitation learning datasets.

By Yuquan Xue, Guanxing Lu, Zhenyu Wu, Chuanrui Zhang, Bofang Jia, Zhengyi Gu, Ziwei Wang

robotics multimodal benchmarks

arXiv AI

2d ago

Rewind-IL: Online Failure Detection and State Respawning for Imitation Learning

arXiv:2604.16683v2 Announce Type: replace-cross Abstract: Imitation learning has enabled robots to acquire complex visuomotor manipulation skills from demonstrations, but deployment failures remain a...

By Gehan Zheng, Sanjay Seenivasan, Matthew Johnson-Roberson, Weiming Zhi

llms robotics multimodal

One email a morning, machine-written

Related stories

Phase-Localized Curation Does Not Help: A Negative Result on Per-Phase Metric Selection for Demonstration Filtering

Rethinking Demonstration Unlearning in Imitation Learning for Robotics

What Demonstration Curation Metrics Do to Your Policy

Auditing Instruction-Trajectory Mismatches in Multimodal Robot Demonstrations

RESample: A Robust Data Augmentation Framework via Exploratory Sampling for Robotic Manipulation

Rewind-IL: Online Failure Detection and State Respawning for Imitation Learning