![[2602.22115] Slice and Explain: Logic-Based Explanations for Neural Networks through Domain Slicing](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.22115] Slice and Explain: Logic-Based Explanations for Neural Networks through Domain Slicing
Summary
The paper presents a novel approach called 'Slice and Explain,' which utilizes domain slicing to enhance the efficiency of logic-based explanations for neural networks, achieving up to 40% reduction in explanation time.
Why It Matters
As neural networks become increasingly prevalent, their lack of interpretability poses challenges in various applications. This research addresses the critical need for efficient and understandable explanations, making AI systems more transparent and trustworthy, which is essential for both developers and end-users.
Key Takeaways
- Introduces domain slicing as a method to improve explanation efficiency for neural networks.
- Demonstrates a significant reduction in explanation time by up to 40%.
- Addresses the scalability concerns associated with logic-based explanation methods.
Computer Science > Logic in Computer Science arXiv:2602.22115 (cs) [Submitted on 25 Feb 2026] Title:Slice and Explain: Logic-Based Explanations for Neural Networks through Domain Slicing Authors:Luiz Fernando Paulino Queiroz, Carlos Henrique Leitão Cavalcante, Thiago Alves Rocha View a PDF of the paper titled Slice and Explain: Logic-Based Explanations for Neural Networks through Domain Slicing, by Luiz Fernando Paulino Queiroz and Carlos Henrique Leit\~ao Cavalcante and Thiago Alves Rocha View PDF HTML (experimental) Abstract:Neural networks (NNs) are pervasive across various domains but often lack interpretability. To address the growing need for explanations, logic-based approaches have been proposed to explain predictions made by NNs, offering correctness guarantees. However, scalability remains a concern in these methods. This paper proposes an approach leveraging domain slicing to facilitate explanation generation for NNs. By reducing the complexity of logical constraints through slicing, we decrease explanation time by up to 40\% less time, as indicated through comparative experiments. Our findings highlight the efficacy of domain slicing in enhancing explanation efficiency for NNs. Comments: Subjects: Logic in Computer Science (cs.LO); Machine Learning (cs.LG) Cite as: arXiv:2602.22115 [cs.LO] (or arXiv:2602.22115v1 [cs.LO] for this version) https://doi.org/10.48550/arXiv.2602.22115 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Relate...
