arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1606.04155cs.CLcs.NE
收藏

实现神经预测

Rationalizing Neural Predictions

Computer Science and Artificial Intelligence Laboratory(计算机科学与人工智能实验室);Massachusetts Institute of Technology(麻省理工学院)

Tao Lei, Regina Barzilay, Tommi Jaakkola

摘要:

没有理由的预测适用性有限。作为一种补救措施,我们学习提取输入文本片段作为理由--基本原理--这些理由被剪裁得简短而连贯,但足以做出相同的预测。我们的方法结合了两个模块化组件,发电机和编码器,这是训练有素的一起运作。生成器指定文本片段上的分布作为候选理由,并将其传递给编码器进行预测。在训练中从来不给小费。相反,该模型是正规化的desiderata的理由。我们评估的方法对多方面的情感分析手动标注的测试用例。我们的方法比基于注意力的基线有很大的优势。我们还成功地说明了问题检索任务的方法。

英文摘要:

Prediction without justification has limited applicability. As a remedy, we learn to extract pieces of input text as justifications -- rationales -- that are tailored to be short and coherent, yet sufficient for making the same prediction. Our approach combines two modular components, generator and encoder, which are trained to operate well together. The generator specifies a distribution over text fragments as candidate rationales and these are passed through the encoder for prediction. Rationales are never given during training. Instead, the model is regularized by desiderata for rationales. We evaluate the approach on multi-aspect sentiment analysis against manually annotated test cases. Our approach outperforms attention-based baseline by a significant margin. We also successfully illustrate the method on the question retrieval task.

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。

↑