arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2006.05103cs.LGstat.ML
收藏

凸神经网络的奇异情况

The Curious Case of Convex Neural Networks

IIIT Hyderabad(印度海得拉巴国际信息技术学院);TCS Research(塔塔咨询服务公司研究院);IIT Kanpur(印度理工学院坎普尔分校)

Sarath Sivaprasad, Ankur Singh, Naresh Manwani, Vineet Gandhi

摘要:

在本文中,我们研究的神经网络的输出是一个凸函数的输入的约束制定。我们证明了凸性约束可以在全连接层和卷积层上强制执行,使其适用于大多数架构。凸性约束包括将权重(对于除了第一层之外的所有层)限制为非负,以及使用非减凸激活函数。尽管很简单,但这些约束对网络的泛化能力有着深远的影响。我们得出了三个有价值的见解:(a)输入输出凸神经网络(IOC-NN)自正则化并减少过拟合问题;(b)尽管受到严重约束,但它们优于基本多层感知器,并实现了与基本卷积架构相似的性能;(c)IOC-NN对训练标签中的噪声具有鲁棒性。我们使用三种不同的神经网络架构对标准图像分类数据集进行了彻底的实验和消融研究,证明了所提出的想法的有效性。

英文摘要:

In this paper, we investigate a constrained formulation of neural networks where the output is a convex function of the input. We show that the convexity constraints can be enforced on both fully connected and convolutional layers, making them applicable to most architectures. The convexity constraints include restricting the weights (for all but the first layer) to be non-negative and using a non-decreasing convex activation function. Albeit simple, these constraints have profound implications on the generalization abilities of the network. We draw three valuable insights: (a) Input Output Convex Neural Networks (IOC-NNs) self regularize and reduce the problem of overfitting; (b) Although heavily constrained, they outperform the base multi layer perceptrons and achieve similar performance as compared to base convolutional architectures and (c) IOC-NNs show robustness to noise in train labels. We demonstrate the efficacy of the proposed idea using thorough experiments and ablation studies on standard image classification datasets with three different neural network architectures.

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。

↑