arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2209.05732cs.LGcs.AI
收藏

Rényi散度深度互学习

Rényi Divergence Deep Mutual Learning

Tencent Security Big Data Lab(腾讯安全大数据实验室);School of Software Engineering, Xi’an Jiaotong University(西安交通大学软件学院);Maynooth University(梅努斯大学)

Weipeng Huang, Junjie Tao, Changbo Deng, Ming Fan, Wenqiang Wan, Qi Xiong, Guangyuan Piao

AI总结:

本文提出用Rényi散度替代KL散度改进深度互学习,理论分析其收敛性,实验证明能持续提升模型泛化性能。

摘要:

本文重新审视了深度互学习(DML),一种简单而有效的计算范式。我们提出使用Rényi散度替代KL散度,其更加灵活且可调,以改进原始DML。这一修改能够在有限的额外复杂度下持续提升性能。我们从理论上分析了所提范式的收敛性质,并证明在非凸优化任务的最坏情况下,使用恒定学习率的随机梯度下降法以$\mathcal{O}(1)$偏差收敛。也就是说,学习将到达附近的局部最优,但会在有界范围内继续搜索,这可能有助于缓解过拟合。最后,我们广泛的实验结果表明,将DML与Rényi散度结合具有优势,可进一步提升模型泛化能力。

英文摘要:

This paper revisits Deep Mutual Learning (DML), a simple yet effective computing paradigm. We propose using Rényi divergence instead of the KL divergence, which is more flexible and tunable, to improve vanilla DML. This modification is able to consistently improve performance over vanilla DML with limited additional complexity. The convergence properties of the proposed paradigm are analyzed theoretically, and Stochastic Gradient Descent with a constant learning rate is shown to converge with $\mathcal{O}(1)$-bias in the worst case scenario for nonconvex optimization tasks. That is, learning will reach nearby local optima but continue searching within a bounded scope, which may help mitigate overfitting. Finally, our extensive empirical results demonstrate the advantage of combining DML and Rényi divergence, leading to further improvement in model generalization.

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。

↑