在 2D 卷积神经网络中减少过拟合的常用方法有哪些？
[英]What are common methods for reducing overfitting in a 2D Convolutional Neural Network?
问题描述
I have built a 2D Convolutional Neural Network that uses CIFAR10 as training and testing data for image classifying.
我已经构建了一个 2D 卷积神经网络，它使用 CIFAR10 作为图像分类的训练和测试数据。 What causes overfitting and what are some best methods for reducing it?是什么导致了过拟合，有哪些最好的方法来减少它？
1 个解决方案
解决方案1
-1 2019-10-19 02:37:30
What causes overfitting: having too many parameters than are needed to represent the transfer function of image to class.导致过度拟合的原因：参数过多，无法表示将图像的 function 传输到 class。
Best method to reduce it in a typical CNN image classifier: Dropout在典型的 CNN 图像分类器中减少它的最佳方法：Dropout
声明:本站的技术帖子网页，遵循CC BY-SA 4.0协议，如果您需要转载，请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.
相关问题
我的卷积神经网络过度拟合 - My Convolutional Neural Network is Overfitting
2D卷积神经网络可以转化为1D卷积神经网络吗？ - Can 2D convolutional neural network be converted into 1D convolutional neural network?
减少卷积神经网络的内存需求 - Reducing memory requirements for convolutional neural network
在没有帮助的情况下如何克服卷积神经网络中的过度拟合？ - How to overcome overfitting in convolutional neural network when nothing helps?
Google JAX 1D 卷积神经网络 - Google JAX 1D convolutional neural network
如何在 2D 卷积神经网络中使用 6x6 滤波器大小而不导致负维度？ - How to use 6x6 filter size in a 2D Convolutional Neural Network without causing negative dimensions?
3D 卷积神经网络输入形状 - 3D Convolutional Neural Network input shape
3D卷积神经网络的实现 - Implementation of a 3D convolutional neural network
卷积神经网络的输出意味着什么 - What does output of Convolutional neural network implies
卷积神经网络新手 - Convolutional neural network Newbie
相关标签