ExplainabilityTransparencyTrustworthinesscs.AI
Front-propagation Algorithm: Explainable AI Technique for Extracting Linear Function Approximations from Neural Networks
Javier Viaña
May 25, 2024
arXiv.org
Key Takeaways
-
Speed advantage for real-time applications: The front-propagation algorithm can extract linear explanations in a single forward pass, making it significantly faster than existing methods like Integrated Gradients or Shapley Values, enabling real-time explainability for deployed AI systems.
-
Simpler explanations without sacrificing accuracy: The technique produces accurate linear function approximations of neural network decisions, offering stakeholders straightforward mathematical explanations rather than complex attribution scores, making AI reasoning more interpretable to non-technical users.
-
Practical implementation ready: The authors developed a software package called
front-propand validated it across three different neural network architectures on benchmark datasets, suggesting the method is ready for practical deployment rather than remaining theoretical. -
Potential for parallel processing: Because the algorithm runs in a single forward pass, it can operate in parallel with deployed models without slowing down predictions, addressing a key practical barrier to implementing explainability in production systems.
Abstract
This paper introduces the front-propagation algorithm, a novel eXplainable AI (XAI) technique designed to elucidate the decision-making logic of deep neural networks. Unlike other popular explainability algorithms such as Integrated Gradients or Shapley Values, the proposed algorithm is able to extract an accurate and consistent linear function explanation of the network in a single forward pass of the trained model. This nuance sets apart the time complexity of the front-propagation as it could be running real-time and in parallel with deployed models. We packaged this algorithm in a software called $\texttt{front-prop}$ and we demonstrate its efficacy in providing accurate linear functions with three different neural network architectures trained on publicly available benchmark datasets.