深度学习基础知识

深度学习入门者必看：25个你一定要知道的概念如果你还不了解深度学习有多么强大，不妨就从这篇文章开始~https://mp.weixin.qq.com/s?__biz=MzIzNjc1NzUzMw==&mid=2247485927&idx=1&sn=606f87b569a0eb7c29f1c2374ff698dd&chksm=e8d3be95dfa43783ddbd577ce8cc7db64c1b02bb55480c1d3f2ba8e06e4b948d1fd6b8e2defc&scene=21#wechat_redirect一文看懂迁移学习：怎样用预训练模型搞定深度神经网络？ - 知乎

如何配置神经网络中的层数和节点数 - 云+社区 - 腾讯云人工神经网络有两个重要的超参数，用于控制网络的体系结构或拓扑：层数和每个隐藏层中的节点数。配置网络时，必须指定这些参数的值。https://cloud.tencent.com/developer/article/1178279

python - Ordering of batch normalization and dropout? - Stack Overflowhttps://stackoverflow.com/questions/39691902/ordering-of-batch-normalization-and-dropout

batch norm、relu、dropout 等的相对顺序和BN、dropout的几个问题和思考_永远飞翔的鸟-CSDN博客_dropout在激活函数之前还是之后总结：BN和dropout一般不同时使用，如果一定要同时使用，可以将dropout放置于BN后面。1.batch norm、relu、dropout 等的相对顺序Ordering of batch normalization and dropout in TensorFlow?在Batch Normalization: Accelerating Deep Network Train...https://blog.csdn.net/m0_37870649/article/details/82025238

machine learning - How to compute accuracy for multi class classification problem and how is accuracy equal to weighted precision? - Cross Validatedhttps://stats.stackexchange.com/questions/306742/how-to-compute-accuracy-for-multi-class-classification-problem-and-how-is-accura

https://blog.csdn.net/weixin_37641832/article/details/104434509https://blog.csdn.net/weixin_37641832/article/details/104434509

model evaluation - What is the difference between the average class accuracy and average instance accuracy? - Cross Validatedhttps://stats.stackexchange.com/questions/423585/what-is-the-difference-between-the-average-class-accuracy-and-average-instance-a一文详尽混淆矩阵、准确率、精确率、召回率、F1值、P-R 曲线、ROC 曲线、AUC 值、Micro-F1 和 Macro-F1_My Blogs-CSDN博客_f和t哪个是对哪个是错文章目录二分类的评价指标一、混淆矩阵与 TP、TN、FP、FN二、准确率、精确率、召回率、F1值三、P-R 曲线四、ROC 曲线、AUC 值五、P-R曲线和ROC曲线有什么区别，如何选择？多分类的评价指标Reference二分类的评价指标对于二分类问题，机器预测的和实际的还是会有所偏差，所以我们引入以下几个概念来评价分类器的优良。一、混淆矩阵与 TP、TN、FP、FN混淆矩阵...https://blog.csdn.net/weixin_37641832/article/details/104434509