arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2003.01204cs.CVcs.LG
收藏

基于Net2Net变换的节能鲁棒累积训练

Energy-efficient and Robust Cumulative Training with Net2Net Transformation

Yale University(耶鲁大学)

Aosong Feng, Priyadarshini Panda

摘要:

深度学习已经在几个计算机视觉任务上达到了最先进的精度。然而,与训练这种深度神经网络相关联的计算和能量要求可能相当高。在本文中,我们提出了一种使用Net2Net变换的累积训练策略,与从头开始训练的模型相比,该策略可以在不造成大的准确性损失的情况下实现训练计算效率。我们首先在原始数据集的一个小子集上训练一个小网络(参数较少),然后使用Net2Net变换逐渐扩展网络,以在数据集的较大子集上递增地进行训练。Net2Net的这种增量训练策略利用了函数保持转换,将知识从之前的每个小网络转移到下一个更大的网络,从而降低了整体训练的复杂性。我们的实验表明,与从头开始训练相比,使用VGG 19以等精度训练TinyImageNet的累积训练在计算复杂性方面降低了约2倍。除了训练效率之外,我们的累积训练策略的一个关键优势是,我们可以在Net2Net扩展期间执行修剪,以获得具有最佳配置的最终网络(与从头开始的传统训练相比,推理计算复杂度降低约0.4倍)。我们还证明,从累积训练获得的最终网络产生更好的泛化性能和噪声鲁棒性。此外,我们还表明,通过累积Net 2Net扩展创建的所有网络的相互推理可以改进对抗性输入检测。

英文摘要:

Deep learning has achieved state-of-the-art accuracies on several computer vision tasks. However, the computational and energy requirements associated with training such deep neural networks can be quite high. In this paper, we propose a cumulative training strategy with Net2Net transformation that achieves training computational efficiency without incurring large accuracy loss, in comparison to a model trained from scratch. We achieve this by first training a small network (with lesser parameters) on a small subset of the original dataset, and then gradually expanding the network using Net2Net transformation to train incrementally on larger subsets of the dataset. This incremental training strategy with Net2Net utilizes function-preserving transformations that transfers knowledge from each previous small network to the next larger network, thereby, reducing the overall training complexity. Our experiments demonstrate that compared with training from scratch, cumulative training yields ~2x reduction in computational complexity for training TinyImageNet using VGG19 at iso-accuracy. Besides training efficiency, a key advantage of our cumulative training strategy is that we can perform pruning during Net2Net expansion to obtain a final network with optimal configuration (~0.4x lower inference compute complexity) compared to conventional training from scratch. We also demonstrate that the final network obtained from cumulative training yields better generalization performance and noise robustness. Further, we show that mutual inference from all the networks created with cumulative Net2Net expansion enables improved adversarial input detection.

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。

↑