BriefGPT.xyz
May, 2023
用最小的训练子集重新标记以改变预测结果
Relabel Minimal Training Subset to Flip a Prediction
HTML
PDF
Jinghan Yang, Lequan Yu
TL;DR
该研究利用扩展影响函数提出了一种有效的识别和重新标记最小训练样本以翻转给定预测的过程，评估模型的韧性，并提供有关训练集内偏差的见解。
Abstract
Yang et al. (2023) discovered that removing a mere 1% of
training points
can often lead to the flipping of a prediction. Given the prevalence of
noisy data
in
→