arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.07679cs.LGcs.AI
收藏

谱门网络

Spectral Gating Networks

Sun Yat-sen University, China(中山大学,中国);Nanyang Technological University, Singapore(南洋理工大学,新加坡)

Jusheng Zhang, Yijia Fan, Kaitong Cai, Jing Yang, Yongsen Zheng, Kwok-Yan Lam, Liang Lin, Keze Wang

AI总结:

谱门网络通过引入紧凑的频谱路径和可学习门控,提升模型在保持稳定性和可扩展性的同时的准确率与效率权衡。

摘要:

门机制无处不在,然而在前馈网络中仍存在一个未被充分研究的问题:如何在不牺牲稳定性和可扩展性的情况下引入富含频率的表达能力?这种张力通过基于样条的Kolmogorov-Arnold网络(KAN)参数化得到了暴露,其中网格细化可能导致参数增长和高维中的脆弱优化。为了提出一种在固定参数和训练预算下保持稳定的方法,将频谱能力注入现有的MLP/FFN层,我们引入了谱门网络(SGN),一种即插即用的频谱重参数化方法。SGN在标准激活路径上增加了紧凑的频谱路径,并引入了可学习的门控,使模型能够从稳定的基底行为开始,并在训练过程中逐步将容量分配给频谱特征。频谱路径通过可训练的随机傅里叶特征(学习的频率和相位)来实例化,取代了基于网格的样条并消除了分辨率依赖性。混合GELU-傅里叶公式进一步提高了优化鲁棒性,同时增强了高频保真度。在视觉、NLP、音频和PDE基准测试中,SGN在可比的计算预算下一致改进了准确率-效率权衡,实现了93.15%的CIFAR-10准确率,并在与样条基KAN变体相比达到11.7倍的推理速度。代码和训练好的模型将被发布。

英文摘要:

Gating mechanisms are ubiquitous, yet a complementary question in feed-forward networks remains under-explored: how to introduce frequency-rich expressivity without sacrificing stability and scalability? This tension is exposed by spline-based Kolmogorov-Arnold Network (KAN) parameterizations, where grid refinement can induce parameter growth and brittle optimization in high dimensions. To propose a stability-preserving way to inject spectral capacity into existing MLP/FFN layers under fixed parameter and training budgets, we introduce Spectral Gating Networks (SGN), a drop-in spectral reparameterization. SGN augments a standard activation pathway with a compact spectral pathway and learnable gates that allow the model to start from a stable base behavior and progressively allocate capacity to spectral features during training. The spectral pathway is instantiated with trainable Random Fourier Features (learned frequencies and phases), replacing grid-based splines and removing resolution dependence. A hybrid GELU-Fourier formulation further improves optimization robustness while enhancing high-frequency fidelity. Across vision, NLP, audio, and PDE benchmarks, SGN consistently improves accuracy-efficiency trade-offs under comparable computational budgets, achieving 93.15% accuracy on CIFAR-10 and up to 11.7x faster inference than spline-based KAN variants. Code and trained models will be released.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑