Knowledge is updated very quickly, and you need to keep learning to keep up with the progress of The Times. For example, Teacher Endar Ng talked about the use of TensorFlow in deep learning class, which he must have learned in recent years, because Google opened the source of TensorFlow in a short time.
Ms. Ng used to use Octave to teach, but stopped using it because: It’s obsolete! ! It’s time to learn something new. \
TensorFlow is a very important part of TensorFlow. It is a very important part of TensorFlow. It is a very important part of TensorFlow.
(Huang Haiguang)
Recommended materials ****
1. TensorFlow section of Ng’s Deep Learning Notes ****
This part can be said to be an introduction to popular science, and after learning it, you will have a certain understanding of TensorFlow, which will be easier to follow (Section 3.11, week 3 of the second course, corresponding to note P247-253, notes can be downloaded on my Github: Introduction to Github).
Note screenshot: ****\
Figure: Screenshot of deep learning notes \
2.TensorFlow2.0 example (github star 34000+) ************
After the release of TensorFlow version 2.0, TF2.0 uses Keras, Eager Execution by default, supports cross-platform, simplified API and so on compared with 1.x version. This update brings TF2.0 closer to PyTorch, and a series of annoying concepts will never return. Recommend a big god to write TF2.0 sample code, recommended reference. * * * * * * * *
Resource Address:
Github.com/aymericdami…
Resources:
This resource is designed to provide easy insight into TensorFlow with examples. It includes notebook and annotated source code for ease of reading.
It is suitable for beginners who want to find a clear and concise example of TensorFlow. In addition to the traditional “primitive” TensorFlow implementation, you can also find the latest TensorFlow API practices (such as the layers, estimator, the dataset,…) . \
Last updated (08/17/2019) : New example added (TF2.0). \
Configuration environment:
Python 3.6 + TensorFlow 1.8+
Resources:
0 – Prerequisites \
- Introduction to Machine Learning
- Introduction to MNIST dataset
Introduction of 1 –
- Hello World(includes notebook and py source code). Very simple example. Learn how to print “Hello World” using TensorFlow.
- Basic operations (including notebook and PY source code). A simple example covering the basic operations of TensorFlow.
- TensorFlow Eager API basics (including notebook and PY source code). Start using TensorFlow’s Eager API.
2 – Basic model
- Linear regression (includes notebook and PY source code). Use TensorFlow to achieve linear regression.
- Linear regression (eager API) (includes notebook and PY source code). Linear regression is implemented using TensorFlow’s Eager API.
- Logistic regression (includes notebook and py source code). Use TensorFlow to achieve Logistic regression.
- Logistic regression (Eager API) (contains notebook and PY source code). TensorFlow’s Eager API is used to achieve Logistic regression.
- Nearest neighbor (contains notebook and py source code). Using TensorFlow to achieve the nearest neighbor algorithm.
- K-means (includes notebook and PY source code). Use TensorFlow to build the K-means classifier.
- Random forest (contains notebook and PY source code). Random forest classifier is constructed by TensorFlow.
- Gradient Vdescribed the Decision Tree (GBDT) (includes notebook and PY source code). Gradient lifting decision tree (GBDT) was constructed by TensorFlow.
- Word2Vec (word embedding) (contains notebook and py source code). Build a word embedding model (Word2Vec) from Wikipedia data using TensorFlow.
3 – Neural network
- Supervised learning
- Simple neural network (includes notebook and PY source code). A simple neural network (such as multi-layer perceptron) is constructed to classify MNIST digital data sets. Raw TensorFlow implementation.
- Simple neural network (Tf.Layers/Estimator API) (contains notebook and PY source code). Use TensorFlow’layers’ and the ‘Estimator ‘API to build a simple neural network (e.g., multi-layer Perceptron) to classify MNIST digital data sets.
- Simple neural network (Eager API) (contains notebook and PY source code). Use the TensorFlow Eager API to build a simple neural network (such as multi-layer perceptron) to classify MNIST digital data sets.
- Convolutional Neural networks (contains notebook and py source code). Convolutional neural network is constructed to classify MNIST digital data sets. Raw TensorFlow implementation.
- Convolutional Neural network (Tf. layers/Estimator API) (contains notebook and PY source code). TensorFlow’ Layers’ and ‘Estimator ‘API were used to construct the convolutional neural network to classify MNIST digital data sets.
- Recursive Neural Network (LSTM) (contains notebook and PY source code). Recursive neural network (LSTM) was constructed to classify MNIST digital data sets.
- Two-way LSTM(contains notebook and PY source code). A bi-directional recursive neural network (LSTM) was constructed to classify MNIST digital data sets.
- Dynamic LSTM(contains notebook and PY source code). A recursive neural network (LSTM) is constructed to perform dynamic calculations to classify sequences of different lengths.
- unsupervised
- Autoencoder (includes notebook and py source code). Build an autoencoder to encode the image into a lower dimension and rebuild it.
- Variational autoencoder (contains notebook and PY source code). Variational automatic encoders (VAE) are constructed to encode noise and generate images.
- GAN (Generative Adversarial Networks) (contains the notebook and PY source code). Build a generative adversarial network (GAN) to generate images from noise.
- DCGAN (Deep Convolutional Generative Adversarial Networks) (contains notebook and PY source code). Deep convolutional generative adversarial network (DCGAN) is constructed to generate images from noise.
4 – tools
- Save and restore the model (including notebook and PY source code). Use TensorFlow to save and restore the model.
- Tensorboard – Graphics and loss visualization (includes notebook and PY source code). Use Tensorboard to visualize the calculation and plot the losses.
- Tensorboard – Advanced visualization (contains notebook and PY source code). Learn more about Tensorboard; Visualizing variables, gradients, etc……
5 – Data management
- Build the image dataset (containing the notebook and py source code). Build your own image dataset from an image folder or dataset file using TensorFlow data queues.
- TensorFlow dataset API(contains notebook and PY source code). The TensorFlow dataset API was introduced to optimize the input data pipeline.
6 – the GPU
- Basic operation of multiple Gpus (including notebook and PY source code). A simple example of introducing multiple Gpus in TensorFlow.
- Train neural networks on multiple Gpus (includes notebook and PY source code). A clear and simple implementation of TensorFlow for training convolutional neural networks on multiple Gpus.
The data set
- Some examples require MNIST data sets for training and testing. The official website: yann.lecun.com/exdb/mnist/
3. The machine learning field: * * based on Scikit – Learn and TensorFlow “* * * * * * * * * * \
* * * *
A good introduction to machine learning and TensorFlow is machine Learning in action: SciKit-Learn and TensorFlow. \
The book is divided into two parts. The first part (chapters 1 to 8) covers the basic theories and algorithms of sciKit-Learn, from linear regression to random forest. The second part (Chapters 9~16) discusses deep learning and the commonly used framework TensorFlow, and teaches you how to use TensorFlow to build and train deep neural networks and convolutional neural networks.
This book is well received by machine learning enthusiasts and can be regarded as an introduction to machine learning, with a rating of 8.2 on Douban.
This book assumes some Python programming experience and familiarity with the major Python scientific libraries, particularly NumPy, Pandas, and Matplotlib.
- The author of this book has published the accompanying source code:
Github.com/ageron/hand…
- A public welfare organization in China translated the original English book into a Markdown file (md file is recommended to be read by Typora), which can be downloaded on Github:
Github.com/apachecn/ha…
In view of the slow download speed in China, this site packages two resources, which can be downloaded in Baidu cloud:
Links:
Pan.baidu.com/s/1jihUZrXb…
Extraction code: 0xye
If harmony, please leave a message.
4. Python Deep Learning ****
Josh Gordon of the TensorFlow team recommends this book, TF2.0 is based on Keras. If you’re new to deep learning, it’s a good place to start. Of course the code in this book needs to be changed, but it’s very simple: \
import keras -> from tensorflow import keras
Copy the code
Written by Francois Chollet, the father of Keras and now a researcher of Artificial intelligence at Google, Python Deep Learning provides a detailed introduction to the exploration and practice of deep learning using Python and Keras, including computer vision, natural language processing, generative modeling, and more. The book contains more than 30 code examples and detailed step-by-step explanations. \
The author posted the code on Github, which covers almost everything in this book. After learning this book, readers will have the ability to build their own deep learning environment, build image recognition models, and generate images and words. But there is one small regret: the code is explained and commented entirely in English, which can be a struggle even for those with good English skills.
We believe that this book and code are the best tools for beginners to get started with deep learning and Keras.
I explained and annotated all the code in Chinese, and downloaded some data sets required by the code (especially the “Cat and Dog War” data set), and localized some of the images, and all the code passed the test. (Please run in file order, there is some correlation before and after the code). \
The following code contains about 80% of the knowledge point of the book, code directory:
2.1: A first look at A neural network
53: reapparating movie reviews (类 类) \
Apparition of newswires
3.7: Predicting House prices
4.4: Underfitting and overfitting \
5.1: Introduction to ConvNets (Convolutional Neural Networks) \
5.2: Using convnets with Small Datasets (Train a convolution on a small dataset from scratch
Using a Pre-trained Convolutional Neural Network
5.4: Visualizing what Convnets Learn
6.1: One-hot Encoding of words or characters (One-hot encoding of words or characters) \
6.1: Using Word Embeddings
6.2: Understanding RNNs
6.3: Advanced Usage of RNNs
6.4: Sequence processing with convnets
8.1: Text Generation with LSTM \
8.2: DeepDream (DeepDream)
8.3: Neural style Transfer
8.4: Generating Images with VAEs
8.5: Introduction to GANs (Generative Adversarial Networks)
Chinese notes and explanations are shown as follows:
Figure: Chinese annotation and explanation of code \
The author’s Github:
Github.com/fchollet/de…
Chinese comment code:
Github.com/fengdu78/ma…
Temporarily think of so much, welcome to add messages. (Huang Haiguang)
Machine learning beginner
The public account created by Dr. Huang Haiguang has more than 21,000 followers on Zhihu, and Github ranks among the top 120 in the world (more than 30,000). This public number is committed to the direction of artificial intelligence science articles, for beginners to provide learning routes and basic information. Original works include: Personal Notes on Machine learning, notes on deep learning, etc.
Past wonderful review \
-
All those years of academic philanthropy. – You’re not alone
-
Conscience recommendation: Introduction to machine learning information summary and learning recommendations \
-
Machine Learning Course Notes and Resources (Github star 12000+, baidu cloud image provided)
-
Ng deep learning notes, videos and other resources (Github standard star 8500+, providing Baidu cloud image)
-
Statistical Learning Methods of Python code implementation (Github 7200+) \
-
Carefully organized and translated mathematical materials related to machine learning
-
Introduction to Deep Learning – Python Deep Learning, annotated version of the original code in Chinese and ebook
-
Word2vec (original translation)\