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 2102.12192cs.LGcs.CR
收藏

用于鲁棒神经网络优化的乘法重加权

Multiplicative Reweighting for Robust Neural Network Optimization

Tel Aviv University(特拉维夫大学)

Noga Bar, Tomer Koren, Raja Giryes

AI总结:

针对神经网络在噪声标签下性能下降的问题,提出将专家建议学习中的乘法权重(MW)用于样本重加权,理论证明其与梯度下降结合的收敛性及一维优势,实证显示在CIFAR-10等数据集上提升了噪声标签下的准确性并影响对抗鲁棒性。

摘要:

神经网络因其强大性能而广泛应用,但在训练时存在噪声标签的情况下性能会下降。受专家建议学习场景的启发——其中乘法权重(MW)更新最近被证明对专家建议中的中度数据损坏具有鲁棒性,我们提出在神经网络优化过程中使用MW对样本进行重加权。我们从理论上确立了该方法与梯度下降结合时的收敛性,并证明了其在一维情况下的优势。然后,我们通过在CIFAR-10、CIFAR-100和Clothing1M数据集上展示MW在存在标签噪声时提高了神经网络的准确性,实证验证了我们对一般情况的发现。我们还展示了该方法对 adversarial robustness 的影响。

英文摘要:

Neural networks are widespread due to their powerful performance. Yet, they degrade in the presence of noisy labels at training time. Inspired by the setting of learning with expert advice, where multiplicative weights (MW) updates were recently shown to be robust to moderate data corruptions in expert advice, we propose to use MW for reweighting examples during neural networks optimization. We theoretically establish the convergence of our method when used with gradient descent and prove its advantages in 1d cases. We then validate empirically our findings for the general case by showing that MW improves neural networks' accuracy in the presence of label noise on CIFAR-10, CIFAR-100 and Clothing1M. We also show the impact of our approach on adversarial robustness.

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。

↑