深度学习-keras & Tensorflow

最新推荐文章于 2022-06-04 19:30:13 发布

gengvvip

最新推荐文章于 2022-06-04 19:30:13 发布

阅读量200

点赞数

本文链接：https://blog.csdn.net/qq_23236081/article/details/113307914

版权

TensorFlow-Serving的使用实战案例笔记

目标：

明确整体框架流程，DNN架构和损失函数、评价指标等
主流模型的搭建，比如RNN/LSTM/WDL/GAN
拿实际数据跑通模型

深度学习理论知识

Keras

https://keras.io/zh/

https://keras-cn.readthedocs.io/en/latest/

# Keras中文文档
https://link.zhihu.com/?target=https%3A//keras.io/zh/
https://www.tensorflow.org/guide/keras/sequential_model
https://keras.io/getting_started/

https://juejin.cn/post/6844903570358140935

Tensorflow 2

官方初学者

https://www.tensorflow.org/tutorials/quickstart/beginner

https://www.tensorflow.org/guide

50题真 • 一文入门TensorFlow2.x

https://zhuanlan.zhihu.com/p/111071013

# TensorFlow 2 快速教程，初学者入门必备

https://link.zhihu.com/?target=https%3A//www.cnblogs.com/shiyanlou/p/11752002.html

# TensorFlow中文社区-首页

https://link.zhihu.com/?target=http%3A//www.tensorfly.cn/

# Github的资源

https://github.com/tensorflow/tensorflow/

Tensorflow官方教程（来自github的最后推荐资料）

https://www.tensorflow.org/responsible_ai

https://www.tensorflow.org/resources/learn-ml/basics-of-machine-learning

https://learning.oreilly.com/p/register/

coursera

https://www.coursera.org/learn/getting-started-with-tensor-flow2

https://www.coursera.org/learn/getting-started-with-tensor-flow2?action=enroll#enroll(过年花26个小时完成)

知识记忆

过拟合问题

https://www.tensorflow.org/tutorials/keras/overfit_and_underfit

深度学习重点考虑的不是模型拟合能力，而是模型泛化能力。

In deep learning, the number of learnable parameters in a model is often referred to as the model's "capacity".

Intuitively, a model with more parameters will have more "memorization capacity" and therefore will be able to easily learn a perfect dictionary-like mapping between training samples and their targets, a mapping without any generalization power, but this would be useless when making predictions on previously unseen data.

Always keep this in mind: deep learning models tend to be good at fitting to the training data, but the real challenge is generalization, not fitting.

To find an appropriate model size, it's best to start with relatively few layers and parameters, then begin increasing the size of the layers or adding new layers until you see diminishing returns on the validation loss.

Dropout使得模型不会过多依赖于个别节点，而是使得尽可能多的节点做训练过程中非常活跃。

可能是最有效、最普遍的防治过拟合的方式了。

组合多种抑制过拟合的办法，比单一方法往往更有效。