当前位置:   article > 正文

深度学习吴恩达课程——编程作业汇总_吴恩达高级学习算法作业

吴恩达高级学习算法作业

 COURSE 1 Neural Networks and Deep Learning

Logistic Regression with a Neural Network mindset

欢迎完成你的第一份编程作业!您将构建一个逻辑回归分类器来识别猫。这项作业将引导你如何用神经网络的思维方式来做这件事,因此也将磨练你对深度学习的直觉。
说明:
不要在代码中使用循环(for/while),除非指令明确要求这样做。
你将学会:
构建学习算法的总体架构,包括:
1.初始化参数
2.计算成本函数及其梯度
3.使用优化算法(梯度下降)
按照正确的顺序将上述三个函数集合到一个主模型函数中。

deeplearning.ai-andrewNG/COURSE 1 Neural Networks and Deep Learning/Week 2/Logistic Regression as a Neural Network/Logistic Regression with a Neural Network mindset.ipynb at master · robbertliu/deeplearning.ai-andrewNG (github.com)icon-default.png?t=N7T8https://github.com/robbertliu/deeplearning.ai-andrewNG/blob/master/COURSE%201%20Neural%20Networks%20and%20Deep%20Learning/Week%202/Logistic%20Regression%20as%20a%20Neural%20Network/Logistic%20Regression%20with%20a%20Neural%20Network%20mindset.ipynb

Python Basics with Numpy (optional assignment) 

本练习向您简要介绍Python。即使您以前使用过Python,这也将帮助您熟悉我们需要的函数。
说明:
您将使用Python 3。
避免使用for循环和while循环,除非你被明确告知这样做。
不要修改某些单元格中的(# GRADED FUNCTION [FUNCTION name])注释。如果你改了这个,你的作业就不会被评分了。包含该注释的每个单元格应该只包含一个函数。
在对函数进行编码之后,运行它下面的单元格来检查结果是否正确。
完成本作业后,你将:
1.能够使用python笔记本
2.能够使用numpy函数和numpy矩阵/向量操作
3.了解“广播”的概念
4.能够向量化代码deeplearning.ai-andrewNG/COURSE 1 Neural Networks and Deep Learning/Week 2/Logistic Regression as a Neural Network/Logistic Regression with a Neural Network mindset.ipynb at master · robbertliu/deeplearning.ai-andrewNG (github.com)icon-default.png?t=N7T8https://github.com/robbertliu/deeplearning.ai-andrewNG/blob/master/COURSE%201%20Neural%20Networks%20and%20Deep%20Learning/Week%202/Logistic%20Regression%20as%20a%20Neural%20Network/Logistic%20Regression%20with%20a%20Neural%20Network%20mindset.ipynb

Planar data classification with one hidden layer

欢迎来到第三周的编程作业。现在是时候构建你的第一个神经网络了,它将有一个隐藏层。您将看到该模型与使用逻辑回归实现的模型之间的巨大差异。
您将学习如何:
1.实现一个具有单个隐藏层的2类分类神经网络
2.使用具有非线性激活函数的单元,如tanh
3.计算交叉熵损失
4.实现向前和向后传播

deeplearning.ai-andrewNG/COURSE 1 Neural Networks and Deep Learning/Week 3/Planar data classification with one hidden layer/Planar data classification with one hidden layer.ipynb at master · robbertliu/deeplearning.ai-andrewNG (github.com)icon-default.png?t=N7T8https://github.com/robbertliu/deeplearning.ai-andrewNG/blob/master/COURSE%201%20Neural%20Networks%20and%20Deep%20Learning/Week%203/Planar%20data%20classification%20with%20one%20hidden%20layer/Planar%20data%20classification%20with%20one%20hidden%20layer.ipynb

 Building your Deep Neural Network: Step by Step

欢迎完成第四周的作业(1/ 2)!您之前已经训练了一个2层神经网络(具有单个隐藏层)。本周,你将构建一个深度神经网络,你想要多少层就有多少层!
-在本手册中,您将实现构建深度神经网络所需的所有功能。
在接下来的作业中,你将使用这些函数来构建一个用于图像分类的深度神经网络。
**完成这项作业后,你将能够:**
-使用非线性单元如ReLU来改进你的模型
-建立一个更深层的神经网络(有超过1个隐藏层)

deeplearning.ai-andrewNG/COURSE 1 Neural Networks and Deep Learning/Week 4/Building your Deep Neural Network - Step by Step/Building your Deep Neural Network - Step by Step.ipynb at master · robbertliu/deeplearning.ai-andrewNG (github.com)icon-default.png?t=N7T8https://github.com/robbertliu/deeplearning.ai-andrewNG/blob/master/COURSE%201%20Neural%20Networks%20and%20Deep%20Learning/Week%204/Building%20your%20Deep%20Neural%20Network%20-%20Step%20by%20Step/Building%20your%20Deep%20Neural%20Network%20-%20Step%20by%20Step.ipynb

Deep Neural Network for Image Classification: Application

当你完成这个,你就完成了第4周的最后一个编程作业,也是本课程的最后一个编程作业!

您将使用在之前的作业中实现的函数来构建一个深度网络,并将其应用于cat与非cat分类。希望您能看到与之前的逻辑回归实现相比,准确度有所提高。

完成本作业后,你将能够:

建立并应用深度神经网络进行监督学习。

deeplearning.ai-andrewNG/COURSE 1 Neural Networks and Deep Learning/Week 4/Deep Neural Network Application_ Image Classification/Deep Neural Network - Application.ipynb at master · robbertliu/deeplearning.ai-andrewNG (github.com)icon-default.png?t=N7T8https://github.com/robbertliu/deeplearning.ai-andrewNG/blob/master/COURSE%201%20Neural%20Networks%20and%20Deep%20Learning/Week%204/Deep%20Neural%20Network%20Application_%20Image%20Classification/Deep%20Neural%20Network%20-%20Application.ipynb

 COURSE 2 Improving Deep Neural Networks Hyperparameter tuning, Regularization and Optimization

 Initialization

欢迎来到“改进深度神经网络”的第一个作业。

训练你的神经网络需要指定一个权重的初始值。选择好的初始化方法将有助于学习。如果您完成了此专门化之前的课程,那么您可能遵循了我们的权重初始化说明,并且到目前为止已经成功了。但是如何为一个新的神经网络选择初始化呢?在本手册中,您将看到不同的初始化如何导致不同的结果。

一个精心选择的初始化可以:

加快梯度下降的收敛速度

增加梯度下降收敛到更低的训练(和泛化)误差的几率。

deeplearning.ai-andrewNG/COURSE 2 Improving Deep Neural Networks Hyperparameter tuning, Regularization and Optimization/week 01/Initialization/Initialization.ipynb at master · robbertliu/deeplearning.ai-andrewNG (github.com)icon-default.png?t=N7T8https://github.com/robbertliu/deeplearning.ai-andrewNG/blob/master/COURSE%202%20Improving%20Deep%20Neural%20Networks%20Hyperparameter%20tuning%2C%20Regularization%20and%20Optimization/week%2001/Initialization/Initialization.ipynb

Regularization

欢迎来到这周的第二个作业。深度学习模型具有如此大的灵活性和容量,如果训练数据集不够大,过拟合可能是一个严重的问题。当然,它在训练集上做得很好,但是学习到的网络不会泛化到它从未见过的新例子!
你将学习:在你的深度学习模型中使用正则化。deeplearning.ai-andrewNG/COURSE 2 Improving Deep Neural Networks Hyperparameter tuning, Regularization and Optimization/week 01/Regularization/Regularization.ipynb at master · robbertliu/deeplearning.ai-andrewNG (github.com)icon-default.png?t=N7T8https://github.com/robbertliu/deeplearning.ai-andrewNG/blob/master/COURSE%202%20Improving%20Deep%20Neural%20Networks%20Hyperparameter%20tuning%2C%20Regularization%20and%20Optimization/week%2001/Regularization/Regularization.ipynb

Gradient Checking

欢迎来到本周最后的作业!

在这个作业中,你将学习实现和使用梯度检查。

你是一个致力于在全球范围内提供移动支付的团队的一员,并被要求建立一个深度学习模型来检测欺诈行为——每当有人进行支付时,你想看看这笔支付是否可能是欺诈行为,比如用户的账户是否被黑客接管了。但是反向传播实现起来相当有挑战性,有时还会有bug。由于这是一个任务关键型应用程序,因此您公司的首席执行官希望真正确定反向传播的实现是正确的。你的首席执行官说:“给我一个证据,证明你的反向传播确实有效!”为了保证这一点,你将使用“梯度检查”。

https://github.com/robbertliu/deeplearning.ai-andrewNG/blob/master/COURSE%202%20Improving%20Deep%20Neural%20Networks%20Hyperparameter%20tuning%2C%20Regularization%20and%20Optimization/week%2001/Gradient%20Checking/Gradient%20Checking.ipynbicon-default.png?t=N7T8https://github.com/robbertliu/deeplearning.ai-andrewNG/blob/master/COURSE%202%20Improving%20Deep%20Neural%20Networks%20Hyperparameter%20tuning%2C%20Regularization%20and%20Optimization/week%2001/Gradient%20Checking/Gradient%20Checking.ipynb

Optimization Methods

到目前为止,您一直使用梯度下降来更新参数并最小化成本。在本手册中,您将学习更高级的优化方法,这些方法可以加快学习速度,甚至可能使您获得更好的成本函数的最终值。拥有一个良好的优化算法可能是等待数天与仅仅几个小时就能获得好结果的区别。

https://github.com/robbertliu/deeplearning.ai-andrewNG/blob/master/COURSE%202%20Improving%20Deep%20Neural%20Networks%20Hyperparameter%20tuning%2C%20Regularization%20and%20Optimization/week%2002/Optimization%20methods.ipynbicon-default.png?t=N7T8https://github.com/robbertliu/deeplearning.ai-andrewNG/blob/master/COURSE%202%20Improving%20Deep%20Neural%20Networks%20Hyperparameter%20tuning%2C%20Regularization%20and%20Optimization/week%2002/Optimization%20methods.ipynb

TensorFlow Tutorial

欢迎来到本周的编程作业。到目前为止,你总是使用numpy来构建神经网络。现在,我们将逐步介绍一个深度学习框架,使您能够更轻松地构建神经网络。机器学习框架,如TensorFlow、PaddlePaddle、Torch、Caffe、Keras等,可以显著加快机器学习的发展。所有这些框架都有大量的文档,您可以随意阅读。在这个作业中,你将学习在TensorFlow中做以下事情:初始化变量启动你自己的会话训练算法实现神经网络编程框架不仅可以缩短你的编码时间,有时还可以执行优化来加快你的代码。

COURSE 4 Convolutional Neural Networks

Convolutional Neural Networks: Step by Step

欢迎来到课程四的第一份作业!在本作业中,您将在numpy中实现卷积层(CONV)和池化层(POOL),包括前向传播和(可选的)后向传播。

声明:本文内容由网友自发贡献,不代表【wpsshop博客】立场,版权归原作者所有,本站不承担相应法律责任。如您发现有侵权的内容,请联系我们。转载请注明出处:https://www.wpsshop.cn/w/weixin_40725706/article/detail/1003180
推荐阅读
相关标签
  

闽ICP备14008679号