arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2207.08501cs.LGcs.AI
收藏

基于扩展Garson算法的可解释深度信度网络自动编码器

Explainable Deep Belief Network based Auto encoder using novel Extended Garson Algorithm

Institute for Development and Research in Banking Technology(银行业技术发展与研究所);University of Hyderabad(海得拉巴大学)

Satyam Kumar, Vadlamani Ravi

摘要:

机器学习中最困难的任务是解释训练好的浅层神经网络。深度神经网络(DNN)在大量任务上提供了令人印象深刻的结果,但通常仍不清楚这种经过训练的深度神经网络如何做出决策。提供特征重要性是在浅层和深层神经网络中使用的最重要和最流行的解释技术。在本文中,我们开发了一种算法,扩展了Garson算法的思想来解释基于深度信念网络的自动编码器(DBNA)。它用于确定DBN中每个输入特征的贡献。它可以用于任何类型的神经网络与许多隐藏层。该方法的有效性进行了测试的分类和回归数据集取自文献。通过这种方法确定的重要功能进行比较,对Wald卡方(\c{hi} 2)。对于4个分类数据集中的2个和5个回归数据集中的2个,我们提出的方法导致识别出质量更好的特征,从而导致与Wald \c{hi} 2相比统计上更显著的结果。

英文摘要:

The most difficult task in machine learning is to interpret trained shallow neural networks. Deep neural networks (DNNs) provide impressive results on a larger number of tasks, but it is generally still unclear how decisions are made by such a trained deep neural network. Providing feature importance is the most important and popular interpretation technique used in shallow and deep neural networks. In this paper, we develop an algorithm extending the idea of Garson Algorithm to explain Deep Belief Network based Auto-encoder (DBNA). It is used to determine the contribution of each input feature in the DBN. It can be used for any kind of neural network with many hidden layers. The effectiveness of this method is tested on both classification and regression datasets taken from literature. Important features identified by this method are compared against those obtained by Wald chi square (\c{hi}2). For 2 out of 4 classification datasets and 2 out of 5 regression datasets, our proposed methodology resulted in the identification of better-quality features leading to statistically more significant results vis-à-vis Wald \c{hi}2.

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。

↑