This is a page that curates AI-related papers published worldwide. All content here is summarized using Google Gemini and operated on a non-profit basis. Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.
Towards a Novel Measure of User Trust in XAI Systems
Created by
Haebom
Author
Miquel Mir o-Nicolau, Gabriel Moy a-Alcover, Antoni Jaume-i-Cap o, Manuel Gonz alez-Hidalgo, Adel Ghazel, Maria Gemma Sempere Campello, Juan Antonio Palmer Sancho
Outline
This paper is about the study of explainable AI (XAI) methodology for improving the reliability of deep learning models. In order to solve the problem of opacity of deep learning models, we propose a new reliability measurement index that combines performance indicators and reliability indicators from an objective perspective. Through three case studies, we demonstrate that it shows improved performance and sensitivity to various scenarios compared to existing methods.
Takeaways, Limitations
•
Takeaways:
◦
Presenting a new measurement index to improve the reliability of XAI systems
◦
Improved performance and sensitivity compared to existing methods
◦
Presenting a method to comprehensively evaluate performance and reliability from an objective perspective
•
Limitations:
◦
Further validation of the generalizability of the presented case study is needed.
◦
Need to review applicability to various XAI methodologies and deep learning models
◦
Need to evaluate the computational complexity and efficiency of the proposed measurement indicators
