一句话：提出用概率抽象解读循环神经网络，以突破简单确定性模型表达力不足导致的解释瓶颈。
神经网络作为端到端方案在对象识别、机器翻译等任务上表现突出，但其黑箱性质使人难以理解和信任其关键决策。为解释神经网络，已有工作尝试从中抽取简单的确定性模型，但由于这类模型表达力有限，结果并不理想，表现为准确率低、可扩展性受限。本文提出通过概率抽象来解释循环神经网络（RNN），其核心动机是使用概率抽象模型而非简单确定性模型来刻画网络行为，从而缓解表达力不足的问题。摘要未给出具体评估结果；根据问题陈述，评估应关注所提出抽象方法在解释准确性、可扩展性等方面能否优于既有简单模型。
Neural networks are becoming a popular tool for solving many real-world problems such as object recognition and machine translation, thanks to its exceptional performance as an end-to-end solution. However, neural networks are complex black-box models, which hinders humans from interpreting and consequently trusting them in making critical decisions. Towards interpreting neural networks, several approaches have been proposed to extract simple deterministic models from neural networks. The results are not encouraging (e.g., low accuracy and limited scalability), fundamentally due to the limited expressiveness of such simple models.
DOI 原文 ·
@inproceedings{paperbot5990,
title = {Towards interpreting recurrent neural networks through probabilistic abstraction},
author = {Guoliang Dong and Jingyi Wang and Jun Sun and Yang Zhang and Xinyu Wang and Ting Dai and Jin Song Dong and Xingen Wang},
booktitle = {ASE},
year = {2020},
doi = {10.1145/3324884.3416592}
}