逻辑门神经网络更易验证，且保持良好性能
Logic Gate Neural Networks are Good for Verification
- 用布尔逻辑门替代乘法，构建可符号化验证的稀疏结构
- 在5个基准数据集上实现全局鲁棒性与公平性验证
- 适合需要可解释性与形式化保证的高风险场景
基于学习的系统在多个领域日益广泛应用，但传统神经网络的复杂性给形式化验证带来巨大挑战。与常规神经网络不同，学习型逻辑门网络（LGNs）以布尔逻辑门替代乘法，形成类似电路网表的稀疏架构，天然更适合符号化验证，同时仍具备良好的预测性能。本文提出一种用于验证LGNs全局鲁棒性与公平性的SAT编码方法。我们在五个基准数据集上评估该方法，包括一个新构建的五分类变体，结果表明LGNs不仅易于验证，且保持强预测能力。
原文摘要 · Abstract (English)
Learning-based systems are increasingly deployed across various domains, yet the complexity of traditional neural networks poses significant challenges for formal verification. Unlike conventional neural networks, learned Logic Gate Networks (LGNs) replace multiplications with Boolean logic gates, yielding a sparse, netlist-like architecture that is inherently more amenable to symbolic verification, while still delivering promising performance. In this paper, we introduce a SAT encoding for verifying global robustness and fairness in LGNs. We evaluate our method on five benchmark datasets, including a newly constructed 5-class variant, and find that LGNs are both verification-friendly and maintain strong predictive performance.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。