Using Inverse Functions to Interpret and Debug Black-Box Artificial Intelligence Models
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
Black-Box AI systems often have high predictive accuracy, although they provide little to no transparency in decision making.This causes debugging and interpretability extremely challenging.This study investigates the relationship between inputs and their respective model-outputs, and if inverse or approximate inverse mapping can be meaningful used to understand the model's behavior deriving information about the model and increasing its overall accuracy.The research presented will provide a foundation for inverse functions in terms of AI and machine learning.Through interpretive examples the approach shows us how output-to-input reconstruction can increase the accuracy of the AI along with identifying anomalies in the AI's prediction and causes of poor decisions.The findings of the paper suggest that utilizing inverse functions can serve as an effective tool in improving interpretability even without access to the internal structure of the model.This research will contribute to a broader field of Artificial Intelligence since it highlights how mathematical concepts can be used to enhance AI systems.
Fields of Study & Subjects
Explainable Artificial Intelligence (XAI)
Adversarial Robustness in Machine Learning
Software Testing and Debugging Techniques
Publication details
- Journal / source
- International Research Journal of Modernization in Engineering Technology and Science
- Publication date
- 28 March 2026
- Language
- EN