Pay Attention via Quantization: Enhancing Explainability of Neural Networks via Quantized Activation
Pay Attention via Quantization: Enhancing Explainability of Neural Networks via Quantized Activation
复制标题
通过量化来关注：通过量化激活增强神经网络的可解释性
DOI：
10.1109/access.2023.3264855
复制
发表时间：
2023
期刊：
影响因子：
3.9
通讯作者：
Hiromitsu Awano
中科院分区：
文献类型：
--
作者：
Yuma Tashiro;Hiromitsu Awano
现代深度学习算法由高度复杂的人工神经网络组成，使得人类极难跟踪其推理过程。随着深度学习在社会上的应用，推理错误造成的人员和经济损失问题日益突出，因此有必要开发方法来解释深度学习算法决策的基础。虽然已经提出了一种基于注意力机制的方法来可视化自动驾驶任务中影响转向角预测的区域，但其解释能力较低。本文主要研究了网络激活值中每个比特的重要性是有偏的(即符号和指数比特的权重比尾数比特更重)，这一点在以往的研究中被忽略了。具体地说，本文对网络激活进行量化，鼓励将重要信息聚合到符号位。此外，我们还引入了一种仅限于符号位的注意机制来提高解释能力。使用Udacity数据集进行的数值实验表明，该方法在删除度量方面获得了更高的曲线下面积(AUC)。
Modern deep learning algorithms comprise highly complex artificial neural networks, making it extremely difficult for humans to track their inference processes. As the social implementation of deep learning progresses, the human and economic losses caused by inference errors are becoming increasingly problematic, making it necessary to develop methods to explain the basis for the decisions of deep learning algorithms. Although an attention mechanism-based method to visualize the regions that contribute to steering angle prediction in an automated driving task has been proposed, its explanatory capability is low. In this paper, we focus on the fact that the importance of each bit in the activation value of a network is biased (i.e., the sign and exponent bits are weighted more heavily than the mantissa bits), which has been overlooked in previous studies. Specifically, this paper quantizes network activations, encouraging important information to be aggregated to the sign bit. Further, we introduce an attention mechanism restricted to the sign bit to improve the explanatory power. Our numerical experiment using the Udacity dataset revealed that the proposed method achieves ahigher area under curve (AUC) in terms of the deletion metric.
DOI：
--
发表时间：
2015-02
期刊：
--
影响因子：
--
作者：
Ke Xu;Jimmy Ba;Ryan Kiros;Kyunghyun Cho;Aaron C. Courville;R. Salakhutdinov;R. Zemel;Yoshua Bengio-Yoshua-Ben
通讯作者：
Ke Xu;Jimmy Ba;Ryan Kiros;Kyunghyun Cho;Aaron C. Courville;R. Salakhutdinov;R. Zemel;Yoshua Bengio-Yoshua-Ben
通过二值化关注：通过激活二值化增强神经网络的可解释性
DOI：
--
发表时间：
2022
期刊：
影响因子：
--
作者：
Yuma Tashiro;Hiromitsu Awano
通讯作者：
Hiromitsu Awano