Embedded Encoder-Decoder in Convolutional Networks Towards Explainable AI
This addresses the need for explainable AI in computer vision by providing a transparent model for researchers and practitioners, though it appears incremental as it builds on existing encoder-decoder and CNN methods.
The paper tackled the problem of making convolutional neural networks explainable by proposing a new model (XCNN) that integrates encoder-decoder networks into a CNN architecture to generate heatmaps without extra post-processing, showing success on datasets like CIFAR-10 and Tiny ImageNet with visual assessment indicating outperformance in class-specific feature representation.
Understanding intermediate layers of a deep learning model and discovering the driving features of stimuli have attracted much interest, recently. Explainable artificial intelligence (XAI) provides a new way to open an AI black box and makes a transparent and interpretable decision. This paper proposes a new explainable convolutional neural network (XCNN) which represents important and driving visual features of stimuli in an end-to-end model architecture. This network employs encoder-decoder neural networks in a CNN architecture to represent regions of interest in an image based on its category. The proposed model is trained without localization labels and generates a heat-map as part of the network architecture without extra post-processing steps. The experimental results on the CIFAR-10, Tiny ImageNet, and MNIST datasets showed the success of our algorithm (XCNN) to make CNNs explainable. Based on visual assessment, the proposed model outperforms the current algorithms in class-specific feature representation and interpretable heatmap generation while providing a simple and flexible network architecture. The initial success of this approach warrants further study to enhance weakly supervised localization and semantic segmentation in explainable frameworks.