Scalable Partial Explainability in Neural Networks via Flexible Activation Functions (Student Abstract)

Sun, Schyler C.; Li, Chen; Wei, Zhuangkun; Tsourdos, Antonios; Guo, Weisi

Abstract

Current state-of-the-art neural network explanation methods (e.g. Saliency maps, DeepLIFT, LIME, etc.) focus more on the direct relationship between NN outputs and inputs rather than the NN structure and operations itself, hence there still exists uncertainty over the exact role played by neurons. In this paper, we propose a novel neural network structure with Kolmogorov-Arnold Superposition Theorem based topology and Gaussian Processes based flexible activation function to achieve partial explainability of the neuron inner reasoning. The model feasibility is verified in a case study on binary classification of the banknotes.

Citation

Sun, S. C., Li, C., Wei, Z., Tsourdos, A., & Guo, W. (2021, February). Scalable Partial Explainability in Neural Networks via Flexible Activation Functions (Student Abstract). Presented at AAAI-21: AAAI Conference on Artificial Intelligence, Online

• Presentation Conference Type: Conference Paper (published)
• Conference Name: AAAI-21: AAAI Conference on Artificial Intelligence
• Start Date: Feb 2, 2021
• End Date: Feb 9, 2021
• Acceptance Date: Apr 15, 2021
• Online Publication Date: May 18, 2021
• Publication Date: May 18, 2021
• Deposit Date: Feb 12, 2025
• Print ISSN: 2374-3468
• Electronic ISSN: 2159-5399
• Peer Reviewed: Peer Reviewed
• Volume: 35
• Issue: 18
• Pages: 15899-15900
• Book Title: Proceedings of the AAAI Conference on Artificial Intelligence
• DOI: https://doi.org/10.1609/aaai.v35i18.17946
• Public URL: https://durham-repository.worktribe.com/output/3479431