.png){kind=icon}
Return
Post hoc visual interpretation using a deep learning-based smooth feature network
.png){kind=icon}
PRE
AI
DOI：10.1007/s00500-023-09430-z.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Interpreting deep learning (DL) models is difficult due to the complexity of their internal representations. Given the inherent lack of interpretability, it is challenging to identify the reasoning behind the model's prediction. This research draws inspiration from the dual-process hypothesis of human cognitive processes, which differentiates between low-level, quick, and opaque processing and high-level, slow, and transparent processing. Based on this idea, the research introduces a novel post hoc interpretability model for visually explaining classification issues. The proposed model is analyzed through extensive experiments, especially in the context of image data classification. The findings indicate that the proposed model outperforms state-of-the-art models with reserved 2.40 pixels and an accuracy of 83.41 in the classification evaluation of significant features. The results demonstrated that the proposed model performs better than state-of-the-art interpretable models. Its improved performance makes image classification trustworthy for users across different domains.
Keywords:
Neural networks
Deep learning
Image classification
Interpretability
Visual explanations
Journal
IF:
2.5
Papers:
1.0W
Citations:
2.1W
Organization
Cited Papers
ShareSaveShareSaveShareSaveShareSavePeeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI).png)
IEEE ACCESS
OAAI
IEEE ACCESSIF3.6
OAAIShareSaveTrend Analysis on Adoption of Virtual and Augmented Reality in the Architecture, Engineering, and Construction IndustryDataOAAI
DataIF0
OAAIShareSave
View more