arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.08074cs.CV
收藏

突破胶囊网络的极限

Pushing the Limits of Capsule Networks

Princeton University(普林斯顿大学)

Prem Nair, Rohan Doshi, Stefan Keselj

摘要:

卷积神经网络使用池化和其他降尺度操作来维持特征检测的平移不变性,但是在它们的架构中,它们不显式地维持特征相对于彼此的位置的表示。这意味着它们不像人类那样以相同的方式表示同一对象在不同方向上的两个实例,因此训练它们通常需要广泛的数据增强和非常深入的网络。Google Brain的一个团队最近发布了一条新闻,试图解决这个问题:胶囊网络。虽然普通的CNN使用表示特征存在的标量输出,但CapsNet使用表示实体存在的矢量输出。我们希望以各种增量方式对CapsNet进行压力测试,以更好地了解它们的性能和表现力。从广义上讲,我们研究的目标是:(1)在类似于MNIST的数据集上测试CapsNets,但在特定的方式上更难,以及(2)探索CapsNets的内部嵌入空间和错误来源。

英文摘要:

Convolutional neural networks use pooling and other downscaling operations to maintain translational invariance for detection of features, but in their architecture they do not explicitly maintain a representation of the locations of the features relative to each other. This means they do not represent two instances of the same object in different orientations the same way, like humans do, and so training them often requires extensive data augmentation and exceedingly deep networks. A team at Google Brain recently made news with an attempt to fix this problem: Capsule Networks. While a normal CNN works with scalar outputs representing feature presence, a CapsNet works with vector outputs representing entity presence. We want to stress test CapsNet in various incremental ways to better understand their performance and expressiveness. In broad terms, the goals of our investigation are: (1) test CapsNets on datasets that are like MNIST but harder in a specific way, and (2) explore the internal embedding space and sources of error for CapsNets.

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。

↑