[论文解读] Explaining Deep Convolutional Neural Networks for Image Classification by Evolving Local Interpretable Model-agnostic Explanations

Bin Wang, Wenbin Pei+2 more|arXiv (Cornell University)|Nov 28, 2022

Explainable Artificial Intelligence (XAI)被引用 4

一句话总结

本文提出E-LIME，一种基于遗传算法的方法，用于为图像分类中的深度卷积神经网络生成局部、可解释、与模型无关的解释。通过用基于适应度的遗传算法替代LIME中计算成本较高的采样过程，E-LIME在ResNet、DenseNet和MobileNet上对ImageNet图像的推理速度提升至一分钟以内，预测置信度最高提升47.59%。

ABSTRACT

Deep convolutional neural networks have proven their effectiveness, and have been acknowledged as the most dominant method for image classification. However, a severe drawback of deep convolutional neural networks is poor explainability. Unfortunately, in many real-world applications, users need to understand the rationale behind the predictions of deep convolutional neural networks when determining whether they should trust the predictions or not. To resolve this issue, a novel genetic algorithm-based method is proposed for the first time to automatically evolve local explanations that can assist users to assess the rationality of the predictions. Furthermore, the proposed method is model-agnostic, i.e., it can be utilised to explain any deep convolutional neural network models. In the experiments, ResNet is used as an example model to be explained, and the ImageNet dataset is selected as the benchmark dataset. DenseNet and MobileNet are further explained to demonstrate the model-agnostic characteristic of the proposed method. The evolved local explanations on four images, randomly selected from ImageNet, are presented, which show that the evolved local explanations are straightforward to be recognised by humans. Moreover, the evolved explanations can explain the predictions of deep convolutional neural networks on all four images very well by successfully capturing meaningful interpretable features of the sample images. Further analysis based on the 30 runs of the experiments exhibits that the evolved local explanations can also improve the probabilities/confidences of the deep convolutional neural network models in making the predictions. The proposed method can obtain local explanations within one minute, which is more than ten times faster than LIME (the state-of-the-art method).

研究动机与目标

• 解决深度卷积神经网络（CNN）在医疗诊断和自动驾驶等关键应用中可解释性差的问题。
• 克服LIME作为当前最先进的局部解释方法所面临的高计算成本和固定超像素要求的局限。
• 开发一种与模型无关、高效且人类可解释的解释方法，适用于任何预训练的CNN。
• 通过演化局部解释来提升模型对正确预测的输出概率，从而提高预测置信度。
• 在ResNet、DenseNet和MobileNet等多种CNN架构上，于ImageNet数据集上验证该方法的有效性与高效性。

提出的方法

• 将输入图像中的超像素编码为二进制向量，表示候选解释。
• 使用遗传算法（GA）演化这些二进制向量，其中每个个体代表一组超像素的子集。
• 定义适应度函数，以最大化深度CNN在演化后的超像素子集上对目标类别的预测概率。
• 通过选择、交叉和变异操作，逐代迭代优化解释群体。
• 通过直接在解释空间中优化，避免扰动采样，从而降低计算成本。
• 仅依赖模型的预测输出而非其内部结构，确保方法的与模型无关性。

实验结果

研究问题

• RQ1 遗传算法能否在不依赖LIME等基于采样的方法的前提下，有效演化出深度CNN的局部、人类可解释的解释？
• RQ2 与LIME相比，所提出的基于GA的方法（E-LIME）是否在保持或提升解释质量的同时显著降低计算成本？
• RQ3 E-LIME能否通过演化解释提升深度CNN在图像分类任务中的置信度（预测概率）？
• RQ4 该方法是否真正具备与模型无关性，并在ResNet、DenseNet和MobileNet等不同CNN架构上实现良好泛化？
• RQ5 在多次运行中，演化出的解释是否具有一致性和稳定性，且能捕捉到有意义、可解释的特征？

主要发现

• E-LIME在每张图像上生成局部解释的时间少于一分钟，比LIME的平均10分钟快了十余倍。
• 在ResNet上，E-LIME将萨莫伊犬图像的预测置信度最高提升了13.44%，虎鲨图像提升了2.86%。
• 在DenseNet上，E-LIME使虎鲨图像的置信度提升了17.76%，萨莫伊犬图像提升了3.36%，基于30次运行的平均结果。
• 在MobileNet上，E-LIME使萨莫伊犬图像的置信度提升了47.59%，虎鲨图像提升了4.23%，秃鹰图像提升了4.10%。
• 30次运行中预测概率的标准差始终较低，表明演化出的解释具有高度稳定性和可靠性。
• 该方法展现出强大的与模型无关性，成功在ResNet、DenseNet和MobileNet上实现相似的性能与速度，验证了其通用性。