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 2205.11432cs.CLcs.LG
收藏

可解释和鲁棒NLI模型的跨层预测逻辑推理

Logical Reasoning with Span-Level Predictions for Interpretable and Robust NLI Models

Imperial College London(伦敦帝国学院);University of Edinburgh(爱丁堡大学);UCL(伦敦大学学院);Queen Mary University of London(伦敦玛丽女王大学)

Joe Stacey, Pasquale Minervini, Haim Dubossarsky, Marek Rei

摘要:

当前的自然语言推理(NLI)模型取得了令人印象深刻的结果,有时在分布测试集上进行评估时表现优于人类。然而,由于已知这些模型从注释伪影和数据集偏差中学习,因此不清楚模型在多大程度上学习NLI任务,而不是从其训练数据中的浅层几何学中学习。我们通过为NLI引入一个逻辑推理框架来解决这个问题,创建基于逻辑规则的高度透明的模型决策。与以前的工作不同,我们表明,可以在不降低预测精度的情况下实现更好的可解释性。我们几乎完全保留了SNLI的性能,同时还确定了负责每个模型预测的确切假设跨度。使用e-SNLI人类解释,我们验证了我们的模型在跨度水平上做出了明智的决策,尽管在训练过程中没有使用任何跨度标签。我们可以通过在训练过程中使用e-SNLI解释来进一步提高模型性能和跨度级决策。最后,我们的模型在减少的数据设置中更鲁棒。当仅使用1,000个示例进行训练时,相对于基线,MNLI匹配和不匹配验证集的分布外性能提高了13%和16%。用较少的观测值进行训练可以进一步改善分布内和分布外的性能。

英文摘要:

Current Natural Language Inference (NLI) models achieve impressive results, sometimes outperforming humans when evaluating on in-distribution test sets. However, as these models are known to learn from annotation artefacts and dataset biases, it is unclear to what extent the models are learning the task of NLI instead of learning from shallow heuristics in their training data. We address this issue by introducing a logical reasoning framework for NLI, creating highly transparent model decisions that are based on logical rules. Unlike prior work, we show that improved interpretability can be achieved without decreasing the predictive accuracy. We almost fully retain performance on SNLI, while also identifying the exact hypothesis spans that are responsible for each model prediction. Using the e-SNLI human explanations, we verify that our model makes sensible decisions at a span level, despite not using any span labels during training. We can further improve model performance and span-level decisions by using the e-SNLI explanations during training. Finally, our model is more robust in a reduced data setting. When training with only 1,000 examples, out-of-distribution performance improves on the MNLI matched and mismatched validation sets by 13% and 16% relative to the baseline. Training with fewer observations yields further improvements, both in-distribution and out-of-distribution.

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。

↑