Binary Spiking Neural Networks as Causal Models

2026-04-29

AI Summary

This work addresses the challenge of unreliable explanations in binary spiking neural networks (BSNNs) caused by interference from irrelevant features during decision-making. To this end, it introduces a novel approach that formally models the spiking activity of BSNNs as a binary causal model and integrates abductive reasoning with SAT/SMT solvers to generate precise, pixel-level explanations that provably exclude irrelevant features. Evaluated on the MNIST dataset, the method demonstrates superior reliability compared to post-hoc explanation techniques such as SHAP, offering formally verifiable justifications for BSNN decisions. By providing explanations with rigorous formal guarantees, this approach significantly enhances the interpretability of BSNN-based models.

Abstract

We provide a causal analysis of Binary Spiking Neural Networks (BSNNs) to explain their behavior. We formally define a BSNN and represent its spiking activity as a binary causal model. Thanks to this causal representation, we are able to explain the output of the network by leveraging logic-based methods. In particular, we show that we can successfully use a SAT as well as a SMT solver to compute abductive explanations from this binary causal model. To illustrate our approach, we trained the BSNN on the standard MNIST dataset and applied our SAT-based and SMT-based methods to finding abductive explanations of the network's classifications based on pixel-level features. We also compared the found explanations against SHAP, a popular method used in the area of explainable AI. We show that, unlike SHAP, our approach guarantees that a found explanation does not contain completely irrelevant features.

Problem

• Binary Spiking Neural Networks
• Causal Models
• Explainable AI
• Abductive Explanations
• Feature Relevance

Innovation

• Binary Spiking Neural Networks
• Causal Models
• Abductive Explanations
• SAT/SMT Solvers
• Explainable AI