arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1807.04511cs.LGstat.ML
收藏

利用特征回放训练神经网络

Training Neural Networks Using Features Replay

University of Pittsburgh(匹兹堡大学)

Zhouyuan Huo, Bin Gu, Heng Huang

摘要:

使用反向传播算法训练神经网络需要将误差梯度顺序地传递通过网络。向后锁定阻止我们并行更新网络层并充分利用计算资源。最近,有几个作品试图解耦和并行的反向传播算法。然而,当神经网络深入时,它们都遭受严重的准确性损失或内存爆炸。为了解决这些具有挑战性的问题,我们提出了一种新的并行目标制定的目标函数的神经网络。在此基础上,我们引入了特征重放算法,并证明了在一定条件下,该算法能保证收敛到非凸问题的临界点。最后,将该方法应用于深度卷积神经网络的训练,实验结果表明,该方法比传统方法具有更快的收敛速度、更低的内存消耗和更好的泛化误差.

英文摘要:

Training a neural network using backpropagation algorithm requires passing error gradients sequentially through the network. The backward locking prevents us from updating network layers in parallel and fully leveraging the computing resources. Recently, there are several works trying to decouple and parallelize the backpropagation algorithm. However, all of them suffer from severe accuracy loss or memory explosion when the neural network is deep. To address these challenging issues, we propose a novel parallel-objective formulation for the objective function of the neural network. After that, we introduce features replay algorithm and prove that it is guaranteed to converge to critical points for the non-convex problem under certain conditions. Finally, we apply our method to training deep convolutional neural networks, and the experimental results show that the proposed method achieves {faster} convergence, {lower} memory consumption, and {better} generalization error than compared methods.

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。

↑