深度学习可解释性内涵和分类

Deep Learning Interpretable Connotation and Classification

近年来,深度学习在多个行业得到了广泛应用,效果显著。深度学习虽然具有数学统计原理基础,但是对于任务知识表征学习尚缺乏明确解释。对深度学习理论研究的缺乏将导致即时可以通过各种训练方法使得模型得到满意输出,但不能解释模型内部究竟如何进行工作才得到有效结果。本文从深度学习可解释性内涵和分类角度出发,阐述了深度学习可解释性,以期有助于其他学者研究。

In recent years, deep learning has been widely used in many industries with remarkable results. Although deep learning is based on mathematical and statistical principles, it lacks a clear explanation for task knowledge representation learning.The lack of theoretical research on deep learning will lead to satisfactory output of the model through various training methods, but it cannot explain how the model works inside to obtain effective results. This paper expounds the interpretability of deep learning from the perspective of its connotation and classification, hoping to help other scholars to study it.