arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2210.16458stat.MLcs.LG
收藏

为加权二元交叉熵重构 van Rijsbergen 的 $F_β$ 度量

Reformulating van Rijsbergen's $F_β$ metric for weighted binary cross-entropy

Satesh Ramdhani

AI总结:

本文通过为 $F_β$ 度量假设统计分布并利用拐点算法确定最优 $β$,将其转化为加权二元交叉熵中的动态惩罚权重,从而在分类任务中同时提升性能与可解释性。

摘要:

将性能度量与基于梯度的损失函数分离并不总能得到最优结果,并且可能遗漏重要的聚合信息。本文研究在可微损失函数之外纳入一个性能度量来指导训练结果。其目标是通过对该性能度量假设统计分布以进行动态加权,从而引导模型性能和可解释性。关注点是 van Rijsbergen 的 $F_β$ 度量——一种用于衡量分类性能的常用选择。通过对 $F_β$ 的分布假设,可以借助动态惩罚权重建立其与标准二元交叉熵之间的中间联系。首先,对 $F_β$ 度量进行重构,以便于假设统计分布,并给出累积密度函数的相应证明。这些概率被用于一个拐点曲线算法中,以寻找最优的 $β$ 或 $β_{opt}$。该 $β_{opt}$ 在所提出的加权二元交叉熵中被用作权重或惩罚。在公开数据上的实验以及基准分析表明,与基线相比,无论类别是不平衡还是平衡,该方法大多能取得更好且更具可解释性的结果。例如,对于存在已知标注错误的 IMDB 文本数据,$F_1$ 分数提升了 14%。结果还揭示了本文推导出的惩罚模型族之间的共性,以及优化中所使用的召回率中心或精确率中心参数的适用性。该方法的灵活性可以增强可解释性。

英文摘要:

The separation of performance metrics from gradient based loss functions may not always give optimal results and may miss vital aggregate information. This paper investigates incorporating a performance metric alongside differentiable loss functions to inform training outcomes. The goal is to guide model performance and interpretation by assuming statistical distributions on this performance metric for dynamic weighting. The focus is on van Rijsbergens $F_β$ metric -- a popular choice for gauging classification performance. Through distributional assumptions on the $F_β$, an intermediary link can be established to the standard binary cross-entropy via dynamic penalty weights. First, the $F_β$ metric is reformulated to facilitate assuming statistical distributions with accompanying proofs for the cumulative density function. These probabilities are used within a knee curve algorithm to find an optimal $β$ or $β_{opt}$. This $β_{opt}$ is used as a weight or penalty in the proposed weighted binary cross-entropy. Experimentation on publicly available data along with benchmark analysis mostly yields better and interpretable results as compared to the baseline for both imbalanced and balanced classes. For example, for the IMDB text data with known labeling errors, a 14% boost in $F_1$ score is shown. The results also reveal commonalities between the penalty model families derived in this paper and the suitability of recall-centric or precision-centric parameters used in the optimization. The flexibility of this methodology can enhance interpretation.

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。

↑