1.1.1 Definition of Supervised Learning: Supervised learning is a machine learning method in which a model makes predictions by learning from labeled training data.

1.1.2 Principle of supervised learning: By corresponding input features and labeled samples, the parameters of the model are trained so that it can correctly predict unknown samples.

1.2.1 Linear regression: Use linear models to model the relationship between input features and output markers.

1.2.2 Logistic regression: used to handle binary classification problems by mapping input features to a probability for classification.

1.2.3 Decision tree: Classification and regression analysis are performed through a tree structure, which is easy to understand and has good interpretability.

1.2.4 Support vector machine: Find an optimal hyperplane in high-dimensional space to solve classification and regression problems.

1.3.1 Image recognition: Training through labeled image data to achieve automatic classification and recognition of pictures.

1.3.2 Natural language processing: Use trained models to perform text classification, sentiment analysis and other tasks.

1.3.3 Forecasting and predictive analysis: Predict and analyze future events by establishing a supervised learning model.

2.1.1 Definition of Unsupervised Learning: Unsupervised learning is a machine learning method in which a model learns from unlabeled data for pattern discovery and data analysis.

2.1.2 Principle of unsupervised learning: By finding hidden structures and patterns in data, learning and analysis can be performed without labeling samples.

2.2.1 Clustering algorithm: Classify similar samples into one category. Commonly used clustering algorithms include K-means clustering and hierarchical clustering.

2.2.2 Association rule mining: Discover association rules in data and use them for market basket analysis, recommendation systems, etc.

2.2.3 Principal component analysis: Convert high-dimensional data into low-dimensional data through linear transformation for data dimensionality reduction and feature extraction.

2.3.1 Classification of unlabeled data: For unlabeled data, classify it through unsupervised learning.

2.3.2 Data visualization and dimensionality reduction: Visually display or reduce the dimensionality of high-dimensional data to better understand the internal structure of the data.

2.3.3 Anomaly detection: Find abnormal data that does not match the normal pattern through unsupervised learning.

3.1.1 Definition of reinforcement learning: Reinforcement learning is a machine learning method in which the model learns the optimal behavior strategy through continuous trials by interacting with the environment.

3.1.2 Principle of reinforcement learning: learn from and optimize strategies to obtain maximum rewards by interacting with the environment in a deterministic environment or a stochastic environment.

3.2.1 Q learning: Select the optimal strategy by calculating the value function of different actions in different states.

3.2.2 Policy gradient: The policy is updated according to the gradient direction of the policy, and is used to deal with the problem of continuous action space.

3.2.3 Deep reinforcement learning: Combining the techniques of deep learning and reinforcement learning, neural networks are used to deal with high-dimensional state and action space problems.

3.3.1 Game intelligence: Let the machine learn to play games, such as AlphaGo’s success in Go.

3.3.2 Robot control: Let the robot learn to perform complex tasks, such as autonomous navigation and operation, by interacting with the environment.

3.3.3 Adaptive system optimization: Optimize the control strategy of the system through reinforcement learning algorithms to maximize performance.