Andrew Ng Neural-networks-deep-learning 课程笔记一
2017-09-06 21:12
519 查看
Week1 Introduction to Deep Learning
Tips:在Andrew Ng的课程中,通常使用列向量构成的矩阵来表示一系列样本,如X.shape=(n_x,m),
n_x为特征数,m为样本数
1.Neural Networks Basics
(1)Binary classification
二分类问题二分类问题中,结果为离散的两个值,通常为
0,1
以“识别图片中是否为猫”为例:
目标是训练一个能够图片中的动物进行“是否为猫”的判断的分类器,是则输出
1,否则输出
0
特征矩阵
在计算机中,图片以如图所示的三个分离的矩阵来存储,分别对应红绿蓝三个颜色通道,为了实现对于图像的识别,我们需要将三个分离的矩阵变换为特征矩阵,即一个代表图片中是否有猫的矩阵。为了构造这个特征矩阵,我们需要将原图片reshape为一个
(64*64*3,1)的矩阵,因为原图片的分辨率为
64*64
(2)Logistic Regression
Logistic Regression是一种机器学习算法,用于输出结果只有两种情况的问题,目标是最小化prediction和training data之间的error。
仍以识别图片中的猫为例:
给出输入矩阵X,logistic regression算法能够计算出图片中有猫的概率,即:
logistics regression算法的参数有:
输入矩阵X
训练标签label(一般为1,0)
权重矩阵W
阈值threshold(用于区分有无猫)
输出
激活函数(Sigmoid,ReLu)
相关文章推荐
- Andrew Ng deeplearning.ai专项课程第四课Convolutional Neural Networks第一周笔记
- Andrew Ng deeplearning.ai专项课程第四课Convolutional Neural Networks第三周笔记
- Andrew Ng deeplearning.ai专项课程第四课Convolutional Neural Networks第二周笔记
- 李菲菲课程笔记:Deep Learning for Computer Vision – Introduction to Convolution Neural Networks
- Neural networks and deep learning学习笔记 -1
- Neural Networks and Deep Learning 学习笔记(八)
- Neural Networks and Deep Learning 学习笔记(十二)
- deeplearning.ai Neural Networks and Deep Learning 笔记
- 论文笔记 Ensemble of Deep Convolutional Neural Networks for Learning to Detect Retinal Vessels in Fundus
- Coursera-Deep Learning Specialization 课程之(二):Improving Deep Neural Networks: -weak1编程作业
- neural networks and deep learning 学习笔记
- Neural Networks and Deep Learning 学习笔记(十)
- Neural Networks and Deep Learning 学习笔记(十一)
- Andrew-Coursera-DL课程-笔记part2-1(Practical aspects of Deep Learning)
- Coursera-Deep Learning Specialization 课程之(二):Improving Deep Neural Networks: -weak2编程作业
- Neural Networks for Machine Learning 课程笔记
- Neural Networks and Deep Learning 学习笔记(三)
- Neural Networks and Deep Learning 学习笔记(四)
- 【学习笔记】WEEK 1_Introduction to Deep Learning_Supervised Learning with Neural Networks
- Coursera-Deep Learning Specialization 课程之(二):Improving Deep Neural Networks: -weak3编程作业