Abstract:Owing to the advancements in artificial intelligence (AI) technology, deep learning has found applications in data analysis and problem-related decision-making processes. However, because of the "black-box" characteristics of deep neural networks, the explanations of decision-making processes are ambiguous. Therefore, model developers face challenges while establishing users' trust in these processes and the associated decision results. Understanding and explaining the decision-making processes implemented by deep learning-based predictive models has importance in various fields. Based on existing interpretation algorithms, this study proposed a novel interpretation algorithm for deep convolutional neural network (CNN) models. The proposed algorithm, known as the neural network interpreter-segmentation recognition and interpretation (NNI-SRI) algorithm, uses the general ideas of segmentation and recognition to interpret models. Compared with other mainstream interpretation algorithms, this algorithm has the following advantages. First, it has a higher model interpretation speed. Second, it can label the positive and negative features observed during the model prediction process. Third, fewer parameters and hardware resources are required because only the forward propagation is used. Furthermore, the proposed algorithm showed better scene adaptability under the premise of ensuring the accuracy of interpretation, which is consistent with the CNN convolution process. Theoretically, the NNI-SRI algorithm can be applied to any model, particularly CNN models, such as InceptionV3, Xception, and ResNet. In this study, the interpretation algorithm was applied to the standard Inception deep network model and the spider sex recognition model developed by us, which achieved excellent results and verified the feasibility of the algorithm. (c) 2022 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Shrinking-Magnifying Approach: New Exploration For Generalizing Neural Networks

Novel Approach for Promoting the Generalization Ability of Neural Networks

Improving Generalization of Deep Neural Networks by Optimum Shifting

Investigation on Improving Generalization of Neural Network

Intelligent gradient amplification for deep neural networks

An Efficient Optimization Method for Improving Generalization Performance of Fuzzy Neural Networks.

Visualizing Deep Networks Using Segmentation Recognition and Interpretation Algorithm

Accelerating Deep Learning with Shrinkage and Recall

A generic shift-norm-activation approach for deep learning

Improving One-Shot NAS with Shrinking-and-Expanding Supernet

A new regularization learning method for improving generalization capability of neural network

Localized Generalization Error Model and Its Application to Architecture Selection for Radial Basis Function Neural Network

Spatial Data Augmentation: Improving the Generalization of Neural Networks for Pansharpening.

Reparameterization through Spatial Gradient Scaling

A Generalized Attention Mechanism to Enhance the Accuracy Performance of Neural Networks

Study on Evaluation Ways of Feed-Forward Neural Networks Generalization Ability

Optimizing Neural Networks in the Equivalent Class Space.

Advancing Spiking Neural Networks towards Multiscale Spatiotemporal Interaction Learning

Convolutional Shrinkage Neural Networks Based Model-Agnostic Meta-Learning for Few-Shot Learning

SGLFormer: Spiking Global-Local-Fusion Transformer with high performance

GA-SAM: Gradient-Strength based Adaptive Sharpness-Aware Minimization for Improved Generalization