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 2103.06583cs.CV
收藏

Norm Loss:An efficient yet effective regularization method for deep神经网络

Preprint: Norm Loss: An efficient yet effective regularization method for deep neural networks

Leiden University(莱顿大学);Honda Research Institute Europe GmbH(本田欧洲研究所)

Theodoros Georgiou, Sebastian Schmitt, Thomas Bäck, Wei Chen, Michael Lew

摘要:

卷积神经网络训练可能会遇到各种问题,如梯度爆炸或消失,基于缩放的权重空间对称和协变量移位。为了解决这些问题,研究人员开发了权重正则化方法和激活归一化方法。在这项工作中,我们提出了一种基于斜流形的权重软正则化方法。所提出的方法使用损失函数,其推动每个权重向量具有接近1的范数,即权重矩阵被平滑地转向所谓的斜流形。我们使用两种最先进的架构,即ResNet和wide-ResNet,在非常流行的CIFAR-10,CIFAR-100和ImageNet 2012数据集上评估了我们的方法。我们的方法引入了可忽略的计算开销,结果表明,它是有竞争力的最先进的,在某些情况下优于它。此外,结果是不太敏感的超参数设置,如批量大小和正则化因子。

英文摘要:

Convolutional neural network training can suffer from diverse issues like exploding or vanishing gradients, scaling-based weight space symmetry and covariant-shift. In order to address these issues, researchers develop weight regularization methods and activation normalization methods. In this work we propose a weight soft-regularization method based on the Oblique manifold. The proposed method uses a loss function which pushes each weight vector to have a norm close to one, i.e. the weight matrix is smoothly steered toward the so-called Oblique manifold. We evaluate our method on the very popular CIFAR-10, CIFAR-100 and ImageNet 2012 datasets using two state-of-the-art architectures, namely the ResNet and wide-ResNet. Our method introduces negligible computational overhead and the results show that it is competitive to the state-of-the-art and in some cases superior to it. Additionally, the results are less sensitive to hyperparameter settings such as batch size and regularization factor.

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。

↑