arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.03535cs.LGcs.NAmath.NAmath.OC
收藏

基于多级镜像下降法的神经网络稀疏训练

Sparse Training of Neural Networks based on Multilevel Mirror Descent

Institute of Mathematics(数学研究所);University of Würzburg(乌尔姆大学);Institute of Mathematics, CAIDAS University of Würzburg(数学研究所,CAIDAS乌尔姆大学)

Yannick Lunk, Sebastian J. Scott, Leon Bungert

AI总结:

本文提出了一种基于线性化Bregman迭代/镜像下降的动态稀疏训练算法,通过交替静态和动态稀疏模式更新来利用自然产生的稀疏性,结合稀疏诱导Bregman迭代与自适应冻结网络结构,以高效探索稀疏参数空间并保持稀疏性。通过多级优化框架保证收敛性,并实验证明该算法在标准基准上能产生高稀疏性和准确性的模型,同时在理论FLOPs数量和训练时间上均有显著提升。

摘要:

我们介绍了一种基于线性化Bregman迭代/镜像下降的动态稀疏训练算法,该算法通过在静态和动态稀疏模式更新之间交替,利用自然产生的稀疏性。关键思想是将稀疏诱导的Bregman迭代与自适应冻结网络结构相结合,以在保持稀疏性的同时高效探索稀疏参数空间。我们通过将方法嵌入多级优化框架中,提供收敛保证。此外,我们实验证明,我们的算法可以在标准基准上产生高度稀疏且准确的模型。我们还显示,与SGD训练相比,理论上的FLOPs数量从标准Bregman迭代的38%减少到我们的方法的6%,同时保持测试精度。我们还显示,当使用稀疏感知的CPU实现时,训练时间可减少约50%。

英文摘要:

We introduce a dynamic sparse training algorithm based on linearized Bregman iterations / mirror descent that exploits the naturally incurred sparsity by alternating between periods of static and dynamic sparsity pattern updates. The key idea is to combine sparsity-inducing Bregman iterations with adaptive freezing of the network structure to enable efficient exploration of the sparse parameter space while maintaining sparsity. We provide convergence guaranties by embedding our method in a multilevel optimization framework. Furthermore, we empirically show that our algorithm can produce highly sparse and accurate models on standard benchmarks. We also show that the theoretical number of FLOPs compared to SGD training can be reduced from 38% for standard Bregman iterations to 6% for our method while maintaining test accuracy.We additionally show a training time reduction by about 50%, when using a sparsity-aware CPU implementation of our method.

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。

↑