arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2004.14500cs.CLcs.LGstat.ML
收藏

句子分类任务的后验校准训练

Posterior Calibrated Training on Sentence Classification Tasks

University of Pittsburgh(匹兹堡大学);Carnegie Mellon University(卡内基梅隆大学);M*Modal(M*Modal公司)

Taehee Jung, Dongyeop Kang, Hua Cheng, Lucas Mentch, Thomas Schaaf

摘要:

大多数分类模型首先预测所有类别的后验概率分布,然后选择具有最大估计概率的类别。然而,在许多设置中,后验概率本身的质量(例如,65%的机会患有糖尿病),比单独的最终预测类提供更可靠的信息。当这些方法被证明校准不良时,迄今为止的大多数修复都依赖于后验校准,这会重新调整预测概率,但通常对最终分类影响不大。在这里,我们提出了一个端到端的训练过程,称为后验校准(PosCal)训练,直接优化的目标,同时最大限度地减少预测和经验后验概率之间的差异,我们表明,PosCal不仅有助于减少校准误差,但也提高了任务性能的惩罚下降的性能,这两个目标。我们的PosCal在GLUE上实现了约2.5%的任务性能增益和16.1%的校准误差减少(Wang等人,2018年,与基线相比。我们在xSLUE上实现了相当的任务性能,校准误差减少了13.2%(Kang和Hovy,2019),但没有超过两阶段校准基线。PosCal训练可以很容易地扩展到任何类型的分类任务,作为正则化项的形式。此外,PosCal的优势在于,它在训练过程中逐步跟踪校准目标所需的统计数据,从而有效利用大型训练集。

英文摘要:

Most classification models work by first predicting a posterior probability distribution over all classes and then selecting that class with the largest estimated probability. In many settings however, the quality of posterior probability itself (e.g., 65% chance having diabetes), gives more reliable information than the final predicted class alone. When these methods are shown to be poorly calibrated, most fixes to date have relied on posterior calibration, which rescales the predicted probabilities but often has little impact on final classifications. Here we propose an end-to-end training procedure called posterior calibrated (PosCal) training that directly optimizes the objective while minimizing the difference between the predicted and empirical posterior probabilities.We show that PosCal not only helps reduce the calibration error but also improve task performance by penalizing drops in performance of both objectives. Our PosCal achieves about 2.5% of task performance gain and 16.1% of calibration error reduction on GLUE (Wang et al., 2018) compared to the baseline. We achieved the comparable task performance with 13.2% calibration error reduction on xSLUE (Kang and Hovy, 2019), but not outperforming the two-stage calibration baseline. PosCal training can be easily extendable to any types of classification tasks as a form of regularization term. Also, PosCal has the advantage that it incrementally tracks needed statistics for the calibration objective during the training process, making efficient use of large training sets.

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。

↑