SymNet:卷积神经网络中的对称滤波器
SymNet: Symmetrical Filters in Convolutional Neural Networks
摘要:
对称性存在于自然界和科学中。在图像处理中,用于空间滤波的核具有某种对称性(例如,Sobel算子、高斯算子、拉普拉斯算子)。人工前馈神经网络中的卷积层通常在没有任何约束的情况下考虑核权重。在本文中,我们建议调查的对称性约束的影响,卷积层的图像分类任务,我们的灵感来自初级视觉皮层和常见的图像处理技术。目标是评估在卷积神经网络(CNN)的整个训练过程中,通过修改反向传播算法期间执行的权重更新,对滤波器实施对称约束的可能程度,并评估性能变化。本文的主要假设是对称约束减少了网络中自由参数的数量,并且能够实现与现代训练方法几乎相同的性能。特别是,我们解决了以下情况:x/y轴对称,点反射和反点反射。的性能进行了评估的四个数据库的图像。结果支持的结论是,虽然随机权重提供了更多的自由度的模型,对称性约束提供了类似的性能水平,同时大大减少了自由参数的模型中。这种方法在整个特征提取过程中需要线性相位特性的相位敏感应用中是有价值的。
英文摘要:
Symmetry is present in nature and science. In image processing, kernels for spatial filtering possess some symmetry (e.g. Sobel operators, Gaussian, Laplacian). Convolutional layers in artificial feed-forward neural networks have typically considered the kernel weights without any constraint. In this paper, we propose to investigate the impact of a symmetry constraint in convolutional layers for image classification tasks, taking our inspiration from the processes involved in the primary visual cortex and common image processing techniques. The goal is to assess the extent to which it is possible to enforce symmetrical constraints on the filters throughout the training process of a convolutional neural network (CNN) by modifying the weight update preformed during the backpropagation algorithm and to evaluate the change in performance. The main hypothesis of this paper is that the symmetrical constraint reduces the number of free parameters in the network, and it is able to achieve near identical performance to the modern methodology of training. In particular, we address the following cases: x/y-axis symmetry, point reflection, and anti-point reflection. The performance has been evaluated on four databases of images. The results support the conclusion that while random weights offer more freedom to the model, the symmetry constraint provides a similar level of performance while decreasing substantially the number of free parameters in the model. Such an approach can be valuable in phase-sensitive applications that require a linear phase property throughout the feature extraction process.
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。