|
所在平台: Coursera |
课程主页: https://www.coursera.org/learn/deep-neural-networks-with-pytorch
课程评论：没有评论
课程名称：使用PyTorch的深度神经网络 课程概述： 本课程将教授如何使用PyTorch开发深度学习模型。课程开始于PyTorch的张量及自动微分功能，随后每个部分将涉及不同模型，首先介绍线性回归和逻辑/softmax回归等基础知识。接下来，将学习前馈深度神经网络、不同激活函数的作用、归一化和丢弃层。课程还将涵盖卷积神经网络和迁移学习，最后讨论其他几种深度学习方法。 学习成果： 完成本课程后，学习者将能够： • 解释并应用深度神经网络和相关机器学习方法的知识 • 掌握如何使用Python库（如PyTorch）进行深度学习应用 • 使用PyTorch构建深度神经网络 课程大纲： 1. **张量与数据集** - 概述张量和数据集，介绍如何分类张量中的数据类型和张量类型，学习1D和2D张量的基础知识，以及简单及偏导数的差异，构建简单的数据集类和图像数据集，并在实验中应用所学。 2. **线性回归** - 学习线性回归的基本概念，如何使用nn.Modules构建自定义模块进行预测，探索state_dict()方法，训练模型、定义数据集及噪声假设，最小化成本并计算损失，理解梯度下降法及其在成本函数上的应用。 3. **PyTorch中的线性回归** - 实现使用PyTorch数据加载器的随机梯度下降，了解批处理技术以提高模型训练效率，比较小批量梯度下降与随机梯度下降，学习收敛率及PyTorch优化模块的应用。 4. **多输入输出线性回归** - 使用线性类在多维空间中执行线性回归，学习模型参数、成本计算及在PyTorch中执行梯度下降，扩展线性回归以处理多个输出。 5. **分类的逻辑回归** - 学习线性分类器和逻辑回归的基础，使用nn.sequential模型在PyTorch中构建神经网络，实施逻辑回归进行预测，并理解与逻辑回归相关的统计概念，如伯努利分布和最大似然估计。 6. **实践项目与最终项目** - 实施最终项目，综合运用所学知识，构建逻辑回归模型以预测《英雄联盟》比赛的结果，利用游戏内的各种统计数据创建强大的预测模型。 这个课程将为希望深入了解深度学习并能够实际应用其理论与技术的学习者提供全面、实用的学习体验。
Name:Tensor and Datasets
Description:This module provides an overview of tensors and datasets. It will cover the appropriate methods to classify the type of data in a tensor and the type of tensor. You will learn the basics of 1D and 2-D tensors and the Numel method. Then you will learn to differentiate simple and partial derivatives. The module lists the different attributes that PyTorch uses in order to calculate a derivative. You will build a simple dataset class and object and a dataset for images. You will apply your learnings in labs and test your concepts in quizzes.
Name:Linear Regression
Description:This module describes linear regression. You will learn about classes, and how to build custom modules using nn.Modules to make predictions. Then you will explore the state_dict() method that returns a python dictionary. Then you will learn how to train the model, define a dataset and the noise assumption. You will further see how to minimize the cost and how to calculate loss using PyTorch. You will understand the Gradient Descent method and how to apply it on the cost function. You will learn to determine the bias and slope using the Gradient Descent method and define the cost surface. You will apply your learnings in labs and test your concepts in quizzes.
Name:Linear Regression PyTorch Way
Description:This module covers implementing stochastic gradient descent using PyTorch’s data loader. Then you will explore batch processing techniques for efficient model training. You will compare Mini-Batch Gradient Descent and Stochastic Gradient Descent. Next, you will learn about Convergence Rate and using PyTorch’s optimization modules. Finally, you will learn the best practices for splitting data to ensure robust model evaluation and how hyperparameters are applied to train data. You will apply your learnings in labs and test your concepts in quizzes.
Name:Multiple Input Output Linear Regression
Description:In this module, you will learn to use the class linear to perform linear regression in multiple dimensions. In addition, you will learn about model parameters and how to calculate cost and perform gradient descent in PyTorch. You will learn to extend linear regression for multiple outputs. You will apply your learnings in labs and test your concepts in quizzes.
Name:Logistic Regression for Classification
Description:In this module, you will learn the fundamentals of linear classifiers and logistic regression. You will learn to use the nn.sequential model to build neural networks in PyTorch. You will implement logistic regression for prediction. The module also covers statistical concepts like Bernoulli Distribution and Maximum Likelihood Estimation underpinning logistic regression. In addition, you will understand and implement the cross entropy loss function. You will apply your learnings in labs and test your concepts in quizzes.
Name:Practice Project and Final Project
Description:In this module, you will implement the final project applying all concepts learned. You will build a logistic regression model aimed at predicting the outcomes of League of Legends matches. Leveraging various in-game statistics, this project will utilize your knowledge of PyTorch, logistic regression, and data handling to create a robust predictive model.
The course will teach you how to develop deep learning models using Pytorch. The course will start with Pytorch's tensors and Automatic differentiation package. Then each section will cover different models starting off with fundamentals such as Linear Regression, and logistic/softmax regression. Followed by Feedforward deep neural networks, the role of different activation functions, normalization and dropout layers. Then Convolutional Neural Networks and Transfer learning will be covered. Finally, several other Deep learning methods will be covered. Learning Outcomes: After completing this course, learners will be able to: • explain and apply their knowledge of Deep Neural Networks and related machine learning methods • know how to use Python libraries such as PyTorch for Deep Learning applications • build Deep Neural Networks using PyTorch
