论文标题
具有对比的预训练和对抗性过滤的强大面向任务的对话生成
Robust Task-Oriented Dialogue Generation with Contrastive Pre-training and Adversarial Filtering
论文作者
论文摘要
数据工件激励机器学习模型通过利用数据中的快捷方式来学习不可转移的概括，并且越来越多的证据表明，数据伪像在最近的自然语言处理基准中实现的强大结果起着作用。在本文中，我们专注于以任务为导向的对话，并研究流行数据集（例如Multiwoz）是否包含此类数据工件。我们发现，通过仅保留频繁的短语在训练示例中，最新模型的性能与经过完整数据训练的变体相比类似，这表明它们利用了这些虚假的相关性来解决该任务。在此激励的情况下，我们提出了一个基于对比的学习框架，以鼓励模型忽略这些线索并专注于学习通用模式。我们还尝试对抗过滤以删除“简单”训练实例，以便该模型专注于从“更难”实例中学习。我们进行了许多概括实验，例如跨域/数据集和对抗测试 - 以评估我们方法的鲁棒性，并发现它的工作原理非常好。
Data artifacts incentivize machine learning models to learn non-transferable generalizations by taking advantage of shortcuts in the data, and there is growing evidence that data artifacts play a role for the strong results that deep learning models achieve in recent natural language processing benchmarks. In this paper, we focus on task-oriented dialogue and investigate whether popular datasets such as MultiWOZ contain such data artifacts. We found that by only keeping frequent phrases in the training examples, state-of-the-art models perform similarly compared to the variant trained with full data, suggesting they exploit these spurious correlations to solve the task. Motivated by this, we propose a contrastive learning based framework to encourage the model to ignore these cues and focus on learning generalisable patterns. We also experiment with adversarial filtering to remove "easy" training instances so that the model would focus on learning from the "harder" instances. We conduct a number of generalization experiments -- e.g., cross-domain/dataset and adversarial tests -- to assess the robustness of our approach and found that it works exceptionally well.
