提出高效学习逻辑门的WARP框架，提升推理速度与模型扩展性。
WARP Logic Neural Networks
- 基于沃尔什松弛的梯度优化框架，直接学习硬件原生逻辑块组合。
- 训练收敛更快，支持更深网络与更高输入维度的逻辑函数。
- 适合需要低延迟、高效率推理的嵌入式AI场景。
快速高效的AI推理日益重要，近期直接学习底层逻辑操作的模型已达到顶尖性能。然而，现有逻辑神经网络存在训练成本高、冗余或依赖近似梯度的问题，限制了可扩展性。为此，我们提出用于概率性逻辑的沃尔什松弛（WARP）神经网络——一种基于梯度的新颖框架，能高效学习硬件原生逻辑块的组合。我们证明WARP对精确学习布尔函数具有最紧凑的参数表示，并且多个先前方法可视为其受限特例。通过引入可学习阈值和残差初始化改进训练，同时利用随机平滑弥合松弛训练与离散逻辑推断之间的差距。实验表明，相比当前最优基线，收敛更快，且能有效扩展至更深架构和更高输入维度的逻辑函数。
原文摘要 · Abstract (English)
Fast and efficient AI inference is increasingly important, and recent models that directly learn low-level logic operations have achieved state-of-the-art performance. However, existing logic neural networks incur high training costs, introduce redundancy or rely on approximate gradients, which limits scalability. To overcome these limitations, we introduce WAlsh Relaxation for Probabilistic (WARP) logic neural networks -- a novel gradient-based framework that efficiently learns combinations of hardware-native logic blocks. We show that WARP yields the most parameter-efficient representation for exactly learning Boolean functions and that several prior approaches arise as restricted special cases. Training is improved by introducing learnable thresholding and residual initialization, while we bridge the gap between relaxed training and discrete logic inference through stochastic smoothing. Experiments demonstrate faster convergence than state-of-the-art baselines, while scaling effectively to deeper architectures and logic functions with higher input arity.
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。