论文标题
了解是什么与反事实示例和梯度监督有所不同
Learning What Makes a Difference from Counterfactual Examples and Gradient Supervision
论文作者
论文摘要
限制深度学习适用性的主要挑战之一是它对学习虚假相关性的敏感性，而不是感兴趣的任务的基本机制。仅通过简单地使用来自同一分布的更多数据来解决所产生的概括。我们提出了一个辅助培训目标，该目标通过利用现有数据集中发现的被忽视的监督信号来提高神经网络的概括能力。我们使用了与不同标签（又称反事实或对比示例）的微小差异示例对，这些示例提供了指示任务的基本因果结构的信号。我们表明，可以在计算机视觉（视觉询问，多标签图像分类）和自然语言处理（情感分析，自然语言推断）中的许多现有数据集中识别此类对。新的培训目标通过成对的反事实示例来确定模型决策功能的梯度。接受此技术训练的模型表明，在分布外测试集上的性能提高了。
One of the primary challenges limiting the applicability of deep learning is its susceptibility to learning spurious correlations rather than the underlying mechanisms of the task of interest. The resulting failure to generalise cannot be addressed by simply using more data from the same distribution. We propose an auxiliary training objective that improves the generalization capabilities of neural networks by leveraging an overlooked supervisory signal found in existing datasets. We use pairs of minimally-different examples with different labels, a.k.a counterfactual or contrasting examples, which provide a signal indicative of the underlying causal structure of the task. We show that such pairs can be identified in a number of existing datasets in computer vision (visual question answering, multi-label image classification) and natural language processing (sentiment analysis, natural language inference). The new training objective orients the gradient of a model's decision function with pairs of counterfactual examples. Models trained with this technique demonstrate improved performance on out-of-distribution test sets.
