arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.11795cs.CLcs.AIcs.LG
收藏

辛普森在NLP训练中的偏差

Simpson's Bias in NLP Training

Rakuten Institute of Technology(乐天技术研究所);Microsoft Research(微软研究院);Rakuten, Inc.(乐天株式会社)

Fei Yuan, Longtu Zhang, Huang Bojun, Yaobo Liang

摘要:

在大多数机器学习任务中,我们通过测量群体级别度量$F(S;M)$来评估给定数据群体$S$上的模型$M$。这样的评估度量$F$的示例包括用于(二进制)识别的精确度/召回率、用于多类分类的F1分数以及用于语言生成的BLEU度量。另一方面,通过在每个学习步骤$t$处优化样本级损失$G(S_t;M)$来训练模型$M$,其中$S_t$是$S$的子集(也称为$S$)。小批量)。$G$的流行选择包括交叉熵损失、骰子损失和重复级别BLEU分数。该范例背后的基本假设是,如果在所有可能的样本上求平均,则样本级损失$G$的平均值应该有效地表示任务的群体级度量$F$,例如,$\mathbb{E}[ G(S_t;M) ] \approx F(S;M)$。在本文中,我们系统地研究了上述假设在几个NLP任务。我们在理论和实验上都表明,样本级损失$G$的一些流行设计可能与任务的真实群体级度量$F$不一致,因此,经过训练以优化前者的模型可能基本上是后者的次优,我们称之为辛普森偏差的现象,由于它与统计学和社会科学中的经典悖论辛普森反转悖论有着深刻的联系。

英文摘要:

In most machine learning tasks, we evaluate a model $M$ on a given data population $S$ by measuring a population-level metric $F(S;M)$. Examples of such evaluation metric $F$ include precision/recall for (binary) recognition, the F1 score for multi-class classification, and the BLEU metric for language generation. On the other hand, the model $M$ is trained by optimizing a sample-level loss $G(S_t;M)$ at each learning step $t$, where $S_t$ is a subset of $S$ (a.k.a. the mini-batch). Popular choices of $G$ include cross-entropy loss, the Dice loss, and sentence-level BLEU scores. A fundamental assumption behind this paradigm is that the mean value of the sample-level loss $G$, if averaged over all possible samples, should effectively represent the population-level metric $F$ of the task, such as, that $\mathbb{E}[ G(S_t;M) ] \approx F(S;M)$. In this paper, we systematically investigate the above assumption in several NLP tasks. We show, both theoretically and experimentally, that some popular designs of the sample-level loss $G$ may be inconsistent with the true population-level metric $F$ of the task, so that models trained to optimize the former can be substantially sub-optimal to the latter, a phenomenon we call it, Simpson's bias, due to its deep connections with the classic paradox known as Simpson's reversal paradox in statistics and social sciences.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑