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 2103.16634cs.CVcs.AIcs.LG
收藏

利用不变性训练深度神经网络

Exploiting Invariance in Training Deep Neural Networks

Amazon Web Services(亚马逊云科技)

Chengxi Ye, Xiong Zhou, Tristan McKinney, Yanfeng Liu, Qinggang Zhou, Fedor Zhdanov

摘要:

受动物视觉系统中两种基本机制的启发,我们引入了一种特征变换技术,该技术在深度神经网络的训练中具有不变性。由此产生的算法需要较少的参数调整,训练良好的初始学习率1.0,并很容易推广到不同的任务。我们使用数据中的局部统计数据来实现尺度不变性,以在不同尺度下对齐相似的样本。为了加速收敛,我们使用从一批数据中提取的全局统计信息来强制GL(n)不变性,使得梯度下降解在基变化下保持不变。分析显示,我们提出的修改占用了底层卷积层5%的计算量。在卷积网络和Transformer网络上进行测试,我们提出的技术需要更少的迭代来进行训练,大大超过所有基线,无缝地工作在小批量和大批量训练上,并适用于不同的计算机视觉和语言任务。

英文摘要:

Inspired by two basic mechanisms in animal visual systems, we introduce a feature transform technique that imposes invariance properties in the training of deep neural networks. The resulting algorithm requires less parameter tuning, trains well with an initial learning rate 1.0, and easily generalizes to different tasks. We enforce scale invariance with local statistics in the data to align similar samples at diverse scales. To accelerate convergence, we enforce a GL(n)-invariance property with global statistics extracted from a batch such that the gradient descent solution should remain invariant under basis change. Profiling analysis shows our proposed modifications takes 5% of the computations of the underlying convolution layer. Tested on convolutional networks and transformer networks, our proposed technique requires fewer iterations to train, surpasses all baselines by a large margin, seamlessly works on both small and large batch size training, and applies to different computer vision and language tasks.

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。

↑