foreword

With the vigorous development of artificial intelligence and machine learning, related algorithms and technologies have been widely used in many industries, and a large number of researchers and personnel from various industries have also invested in the research and development of machine learning.

Mastering the principles of advanced machine learning algorithms and being able to flexibly use them according to different situations is an essential core skill for relevant practitioners. It can also help them improve their theoretical level and achieve unique creative results.

Here is a book to recommend to everyone: "PyTorch Advanced Machine Learning Practice "

This book mainly introduces the classic algorithm content in the field of machine learning, as well as the basic knowledge involved in related principles. This part of the content generally appears in the advanced courses at the graduate level, and is the necessary knowledge for in-depth study of machine learning. At the same time, a major feature of this book is that it not only stays at the level of simple introduction to theoretical algorithms, but also emphasizes hands-on practice. For the convenience of readers, this book uses PyTorch, the most popular machine learning framework at present, to realize all algorithm processes. PyTorch has been used in the field of deep learning before, and can realize the training and operation of deep neural networks . This book uses its complete scientific calculation matrix library, flexible automatic differentiation and derivation engine, and convenient GPU accelerated calculations to show readers that the PyTorch framework is also widely used in the field of machine learning.

book catalog

Chapter 1 Overview of Machine Learning/
1.1 Introduction to Machine Learning/
1.1.1 The Meaning of Machine Learning/
1.1.2 Overview of Machine Learning/
1.1.3 Different Types of Machine Learning Algorithms/
1.2 Data Processing/
1.2.1 Data Feature Classification and Representation/
1.2.2 Data preprocessing/
1.2.3 Data missing processing/
1.2.4 Feature derivation and crossover/
1.2.5 Feature screening/
1.3 Metrics/
1.3.1 Model evaluation indicators/
1.3.2 Data set division/
1.3.3 Super Parameter Optimization /
1.4 Optimization Objective /
1.4.1 Loss Function /
1.4.2 Gradient Descent Optimization /
1.4.3 Constrained Optimization: Lagrange Function /
1.5 Practice: Simple Model Realizes Titanic Passenger Survival Probability Prediction /
1.5.1 Problem Description and Data Characteristics /
1.5.2 Simple attribute classification model realizes prediction /
Chapter 2 Introduction to basic operations of PyTorch /
2.1 Introduction to PyTorch /
2.2 Core concepts: Tensor /
2.2.1 Basic operations of Tensor /
2.2.2 Basic mathematical operations /
2.2.3 Index sharding operations /
2.2.4 Class member methods /
2.3 Automatic derivation (Autograd) /
2.3.1 Differentiable tensor/
2.3.2 Function: the basis for automatic differentiation/
2.4 Neural network core module: torch.nn/
2.4.1 Overview of nn.Module/
2.4.2 Functional operation nn.functional/
2.5 Optimizer (optimizer) /
2.5.1 Overview of optimizer /
2.5.2 Learning rate adjustment /
2.5.3 Introduction to classic optimizers /
2.6 Data loading /
2.6.1 Introduction to Dataset and DataLoader /
2.6.2 Preprocessing transformation torchvision.transforms /
2.7 Advanced operations /
2.7.1 GPU Computing /
2.7.2 Using C++ to Implement Custom Operators /
2.8 Practice: Wide & Deep Model Realizes Criteo Click Rate Estimation /
2.8.1 Problem Definition and Data Characteristics /
2.8.2 Wide & Deep Model Introduction /
2.8.3 Complete Experiment Process /
Chapter 3 Supervised Learning /
3.1 Linear Regression /
3.1.1 Least Square Method /
3.1.2 Ridge Regression /
3.1.3 Bayesian Linear Regression (Bayesian Linear Regression) /
3.2 Logistic Regression (Logistic Regression)/
3.2.1 Binary Classification Logistic Regression/
3.2.2 Multi-Classification Softmax Regression/
3.2.3 Bayesian Logistic Regression (Bayesian Logistic Regression)/
3.3 Support Vector Machine (SVM)/
3.3.1 Definition of SVM under linear separability/
3.3.2 Solving using stochastic gradient descent/
3.3.3 Introduction to convex optimization/
3.3.4 SVM dual problem representation/
3.3.5 Gradient descent method to solve dual problems/
3.3.6 From Hard SVM extended to Soft SVM/
3.3.7 Support Vector Regression (SVR)/
3.3.8 SVR with slack variables and dual optimization method/
3.4 Decision Tree model (Decision Tree)/
3.4.1 Building a single tree model /
3.4.2 Ensemble Learning (Ensemble Learning) /
3.5 K Nearest Neighbors (KNN) /
3.6 Actual Combat: Complex Model Realizes Titanic Passenger Survival Probability Prediction /
3.6.1 Titanic Dataset Feature Processing /
3.6.2 Multiple Models Prediction Performance Comparison /
Chapter 4 Unsupervised Learning /
4.1 Clustering Method /
4.1.1 KMeans Clustering /
4.1.2 Spectral Clustering /
4.1.3 Agglomerative Clustering /
4.2 Density Estimation /
4.2.1 Gaussian Mixture Model /
4.2 .2 Expectation Maximization (EM) /
4.3 Dimension Reduction & Embedding (Dimension Reduction & Embedding) /
4.3.1 Principal Component Analysis (PCA) /
4.3.2 Locally Linear Embedding (Locally Linear Embedding, LLE)/
4.3.3 Stochastic Neighbor Embedding Algorithm (t-SNE)/
4.4 Practice: Unsupervised Method for Anomaly Detection/
4.4.1 Anomaly Detection Problems and Applications/
4.4.2 Realization of PCA-Based Anomaly Detection Method/
4.4.3 Implementation of anomaly detection method based on Mahalanobis distance /
4.4.4 Implementation of local anomaly factor detection method based on clustering /
Chapter 5 PyTorch advanced machine learning combat probability graph model /
5.1 Directed graph: Bayesian Network )/
5.1.1 Probabilistic Decomposition of Directed Graphs/
5.1.2 Conditional Independence/
5.2 Undirected Graph: Markov Random Field (MRF) /
5.2.1 Probability Decomposition of Undirected Graph /
5.2.2 Specific Application: Image Denoising /
5.3 Hidden Markov Model (Hidden Markov Model, HMM)/
5.3.1 Hidden Markov Model Introduction/
5.3.2 Forward-Backward Algorithm (Forward-Backward Algorithm)/
5.3.3 Scaling to Improve Calculation Stability/
5.3.4 Code Implementation/
5.4 Variation Inference (Variational Inference, VI) /
5.4.1 Posterior Distribution Optimization and ELBO /
5.4.2 Black-Box Variational Inference Algorithm (BBVI) /
5.5 Monte Carlo Sampling /
5.5. 1 Rejection Sampling /
5.5.2 Markov Chain Monte Carlo /
5.5.3 Gibbs Sampling /
5.5.4 Hamiltonian Monte Carlo , HMC) /
5.6 Practice: Variational Gaussian Mixture Model /
5.6.1 Extended GMM: Bayesian Gaussian Mixture Model /
5.6.2 Variational Inference Approximation /
5.6.3 Code Implementation /
Chapter 6 Kernel Method /
6.1 Kernel Function and Kernel Technique /
6.2 Kernel KMeans Algorithm (Kernel KMeans) /
6.2.1 Review of KMeans Algorithm /
6.2.2 Specific Implementation /
6.3 Kernelized Support Vector Machine (Kernel SVM) /
6.3.1 SVM Dual Problem and Kernel Function Representation /
6.3.2 Kernelized Support Vector Regression (Kernel SVR) /
6.4 Kernelized Principal Component Analysis (Kernel PCA, KPCA) /
6.4. 1 Review of PCA and Kernelization Representation/
6.4.2 Core Centralization Skills and Implementation/
6.5 Gaussian Process (GP)/
6.5.1 Definition and Basic Properties of Gaussian Process/
6.5.2 Selection and Optimization of Kernel Function Parameters/
6.6 Practice: Hyperparameter optimization using Gaussian process /
6.6.1 Hyperparameter Optimization /
6.6.2 Implementation /
Chapter 7 Deep Neural Network /
7.1 Neural Network /
7.1.1 Basic operator operations /
7.1.2 Common network structure/
7.1.3 Network training/
7.2 Variational AutoEncoder (VAE)/
7.2.1 Introduction to Various Autoencoders/
7.2.2 Variational Autoencoders/
7.3 Deep Generative Model (DGM)/
7.3.1 Restricted Boltzmann Machine (Restricted Boltzmann Machine, RBM)/
7.3 .2 Generative Adversarial Network (GAN) /
7.4 Actual Combat: Image Style Transformation Using CycleGAN /
7.4.1 Introduction to CycleGAN Model /
7.4.2 Model Implementation /
Chapter 8 Reinforcement Learning /
8.1 Introduction to Classical Reinforcement Learning /
8.1 .1 Introduction to basic concepts/
8.1.2 Reinforcement learning environment OpenAI Gym/
8.2 Markov decision process (Markov Decision Process, MDP)/
8.2.1 MDP definition and Bellman superiority equation/
8.2.2 Policy iteration (Policy Iteration) and Value Iteration /
8.2.3 Monte Carlo Learning /
8.2.4 Temporal Difference Learning (TDLearning) /
8.3 Deep Reinforcement Learning Based on Q-Value Function /
8.3.1 Deep Q-Network (DQN) /
8.3.2 Other improved DQN models /
8.4 Deep reinforcement learning based on policy optimization /
8.4.1 Policy Gradient algorithm (Policy Gradient) /
8.4.2 Advantage ActorCritic (A2C) algorithm /
8.4.3 Proximal Policy Optimization (PPO) /
8.4.4 Deep Deterministic Policy Gradient (DDPG) /
8.4.5 Soft Actor Critic (SAC) Algorithm /
8.5 Practice: Deep Reinforcement Learning Evaluation in Atari Game Environment /
8.5.1 Atari Game Environment and Preprocessing Methods /
8.5.2 Performance Comparison of Various Deep Reinforcement Learning /

Lottery

The old rules of giving books by lottery (no liking, collection and winning are invalid): Pay attention and remember to follow the blogger, otherwise you won’t know if you won the lottery! ! !

• 1. Like and collect articles
• 2. Leave a message in the comment area: Life is too short, I use Python! ! ! (Leave a message to enter the prize pool, and each person can leave a maximum of three messages)
• 3. There are 5 people in the crawler draw at 8:00 on Sunday
• If you don’t want to draw a lottery, JD’s self-operated purchase link: https://item.jd.com/13841860.html?