# RustyML
A comprehensive machine learning and deep learning library written in pure Rust.
一个用纯Rust编写的全面机器学习和深度学习库。
[](https://www.rust-lang.org/)
[](https://github.com/SomeB1oody/RustyML/blob/master/LICENSE)
## Overview | 概述
RustyML aims to be a feature-rich machine learning and deep learning framework that leverages Rust's performance, memory safety, and concurrency features. While currently in early development stages, the project's long-term vision is to provide a complete ecosystem for machine learning, deep learning, and transformer-based models.
RustyML 旨在成为一个功能丰富的机器学习和深度学习框架，充分利用Rust的性能、内存安全性和并发特性。虽然目前处于早期开发阶段，但项目的长期愿景是提供一个完整的机器学习、深度学习和基于transformer架构的模型生态系统。
## Key Features | 核心特性
- **Rich Algorithm Collection** | 丰富的算法集合: Supervised, unsupervised learning, and neural networks | 监督学习、无监督学习和神经网络
- **Comprehensive Metrics** | 全面的评估指标: Evaluation tools for regression, classification, and clustering | 回归、分类和聚类的评估工具
- **Model Persistence** | 模型持久化: Save and load trained models with JSON serialization | 通过JSON序列化保存和加载训练好的模型
## Architecture | 架构
### Machine Learning | 机器学习 (`features = ["machine_learning"]`)
Classical machine learning algorithms for supervised and unsupervised learning:
经典机器学习算法，用于监督和无监督学习：
- **Regression | 回归**:
- Linear Regression with L1/L2 regularization | 带L1/L2正则化的线性回归
- **Classification | 分类**:
- Logistic Regression | 逻辑回归
- KNN (K-Nearest Neighbors) | K近邻
- Decision Tree (ID3, C4.5, CART) | 决策树（ID3、C4.5、CART）
- SVC (Support Vector Classification) | 支持向量分类
- Linear SVC | 线性支持向量分类
- LDA (Linear Discriminant Analysis) | 线性判别分析
- **Clustering | 聚类**:
- KMeans with K-means++ initialization | K均值（带K-means++初始化）
- DBSCAN (Density-based clustering) | DBSCAN（基于密度的聚类）
- MeanShift | 均值漂移
- **Anomaly Detection | 异常检测**:
- Isolation Forest | 隔离森林
### Neural Network | 神经网络 (`features = ["neural_network"]`)
Complete neural network framework with flexible architecture design:
完整的神经网络框架，具有灵活的架构设计：
- **Layers | 层**:
- Dense: Fully connected layers with customizable activation functions | Dense层: 具有可定制激活功能的全连接层
- Activation: Standalone activation functions (ReLU, Sigmoid, Tanh, Softmax, etc.) | 激活函数: 独立激活函数（ReLU, Sigmoid, Tanh， Softmax等）
- Pooling Layers: Max pooling and average pooling operations for 1D, 2D, and 3D data | 池化层: 对1D、2D和3D数据进行最大池化和平均池化操作
- Global Pooling: Global max pooling and global average pooling for 1D, 2D, and 3D tensors | 全局池化层: 对1D、2D和3D张量进行全局最大池化和全局平均池化
- Recurrent Layers: Sequential Modeling Layers like RNN, LSTM, and GRU | 循环层: 序列建模层，如RNN, LSTM和GRU
- Dropout: Regularization layer to prevent overfitting during training | Dropout层：正则化层，防止训练过程中的过拟合
- **Optimizers | 优化器**:
- SGD (Stochastic Gradient Descent) | 随机梯度下降
- Adam | Adam优化器
- RMSProp | RMSProp优化器
- AdaGrad | AdaGrad优化器
- **Loss Functions | 损失函数**:
- MSE (Mean Squared Error) | 均方误差
- MAE (Mean Absolute Error) | 平均绝对误差
- Binary Cross-Entropy | 二元交叉熵
- Categorical Cross-Entropy | 分类交叉熵
- Sparse Categorical Cross-Entropy | 稀疏分类交叉熵
- **Models | 模型**:
- Sequential architecture for feed-forward networks | 用于前馈网络的顺序架构
- **Activation layers | 激活层**:
- ReLU, Tanh, Sigmoid, Softmax
### Utility | 工具 (`features = ["utility"]`)
Data preprocessing and dimensionality reduction utilities:
数据预处理和降维工具：
- **Dimensionality Reduction | 降维**:
- PCA (Principal Component Analysis) | 主成分分析
- Kernel PCA (with RBF, Linear, Polynomial, Sigmoid kernels) | 核主成分分析（支持RBF、线性、多项式、Sigmoid核）
- LDA (Linear Discriminant Analysis) | 线性判别分析
- t-SNE (t-Distributed Stochastic Neighbor Embedding) | t-分布随机邻域嵌入
- **Preprocessing | 预处理**:
- Standardization (z-score normalization) | 标准化（z分数归一化）
- Train-test splitting | 训练测试集分割
- **Kernel Functions | 核函数**:
- RBF, Linear, Polynomial, Sigmoid | RBF、线性、多项式、Sigmoid
### Metric | 评估指标 (`features = ["metric"]`)
Comprehensive evaluation metrics for model performance assessment:
用于模型性能评估的全面评估指标：
- **Regression Metrics | 回归指标**:
- MSE (Mean Squared Error) | 均方误差
- RMSE (Root Mean Squared Error) | 均方根误差
- MAE (Mean Absolute Error) | 平均绝对误差
- R² score | R²分数
- **Classification Metrics | 分类指标**:
- Accuracy | 准确率
- Confusion Matrix (with TP, FP, TN, FN, precision, recall, F1-score) | 混淆矩阵（包含TP、FP、TN、FN、精确率、召回率、F1分数）
- AUC-ROC | AUC-ROC曲线下面积
- **Clustering Metrics | 聚类指标**:
- Adjusted Rand Index (ARI) | 调整兰德指数
- Normalized Mutual Information (NMI) | 标准化互信息
- Adjusted Mutual Information (AMI) | 调整互信息
- Silhouette Score | 轮廓系数
### Math | 数学工具 (`features = ["math"]`)
Mathematical utilities and statistical functions:
数学工具和统计函数：
- **Distance Metrics | 距离度量**:
- Euclidean distance | 欧几里得距离
- Manhattan distance | 曼哈顿距离
- Minkowski distance | 闵可夫斯基距离
- **Impurity Measures | 不纯度度量**:
- Entropy | 熵
- Gini impurity | 基尼不纯度
- Information Gain | 信息增益
- Gain Ratio | 增益率
- **Statistical Functions | 统计函数**:
- Variance | 方差
- Standard deviation | 标准差
- SST (Sum of Squares Total) | 总平方和
- SSE (Sum of Squared Errors) | 误差平方和
- **Activation Functions | 激活函数**:
- Sigmoid | Sigmoid函数
- Logistic loss | 逻辑损失
### Dataset | 数据集 (`features = ["dataset"]`)
Access to standardized datasets for experimentation:
用于实验的标准化数据集：
- Boston Housing | 波士顿房价数据集
- Wine Quality (red and white wines) | 葡萄酒质量数据集（红葡萄酒和白葡萄酒）
- Titanic | 泰坦尼克号数据集
## Getting Started | 开始使用
### Machine Learning Example | 机器学习示例
Add the library to your `Cargo.toml`:
将库添加到您的`Cargo.toml`文件中：
```toml
[dependencies]
rustyml = {version = "*", features = ["machine_learning"]}
# or use `features = ["full"]` to enable all features
# Or use features = ["default"] to enable default modules (`machine_learning` and `neural_network`)
```
In your Rust code, write:
在你的Rust代码里写：
``` rust
use rustyml::machine_learning::linear_regression::*;
use ndarray::{Array1, Array2};
// Create a linear regression model
let mut model = LinearRegression::new(true, 0.01, 1000, 1e-6, None).unwrap();
// Prepare training data
let raw_x = vec![vec![1.0, 2.0], vec![2.0, 3.0], vec![3.0, 4.0]];
let raw_y = vec![6.0, 9.0, 12.0];
// Convert Vec to ndarray types
let x = Array2::from_shape_vec((3, 2), raw_x.into_iter().flatten().collect()).unwrap();
let y = Array1::from_vec(raw_y);
// Train the model
model.fit(&x.view(), &y.view()).unwrap();
// Make predictions
let new_data = Array2::from_shape_vec((1, 2), vec![4.0, 5.0]).unwrap();
let _predictions = model.predict(&new_data.view());
// Save the trained model to a file
model.save_to_path("linear_regression_model.json").unwrap();
// Load the model from the file
let loaded_model = LinearRegression::load_from_path("linear_regression_model.json").unwrap();
// Use the loaded model for predictions
let _loaded_predictions = loaded_model.predict(&new_data.view());
// Since Clone is implemented, the model can be easily cloned
let _model_copy = model.clone();
// Since Debug is implemented, detailed model information can be printed
println!("{:?}", model);
```
### Neural Network Example | 神经网络示例
Add the library to your `Cargo.toml`:
将库添加到您的`Cargo.toml`文件中：
```toml
[dependencies]
rustyml = {version = "*", features = ["neural_network"]}
# or use `features = ["full"]` to enable all features
# Or use `features = ["default"]` to enable default modules (`machine_learning` and `neural_network`)
```
In your Rust code, write:
在你的Rust代码里写：
``` rust
use rustyml::neural_network::{
sequential::Sequential,
layer::{Dense, ReLU, Softmax},
optimizer::Adam,
loss_function::CategoricalCrossEntropy,
};
use ndarray::Array;
// Create training data
let x = Array::ones((32, 784)).into_dyn(); // 32 samples, 784 features
let y = Array::ones((32, 10)).into_dyn(); // 32 samples, 10 classes
// Build a neural network
let mut model = Sequential::new();
model
.add(Dense::new(784, 128, ReLU::new()).unwrap())
.add(Dense::new(128, 64, ReLU::new()).unwrap())
.add(Dense::new(64, 10, Softmax::new()).unwrap())
.compile(Adam::new(0.001, 0.9, 0.999, 1e-8).unwrap(), CategoricalCrossEntropy::new());
// Display model structure
model.summary();
// Train the model
model.fit(&x, &y, 10).unwrap();
// Save model weights to file
model.save_to_path("model.json").unwrap();
// Create a new model with the same architecture
let mut new_model = Sequential::new();
new_model
.add(Dense::new(784, 128, ReLU::new()).unwrap())
.add(Dense::new(128, 64, ReLU::new()).unwrap())
.add(Dense::new(64, 10, Softmax::new()).unwrap());
// Load weights from file
new_model.load_from_path("model.json").unwrap();
// Compile before using (required for training, optional for prediction)
new_model.compile(Adam::new(0.001, 0.9, 0.999, 1e-8).unwrap(), CategoricalCrossEntropy::new());
// Make predictions with loaded model
let predictions = new_model.predict(&x);
println!("Predictions shape: {:?}", predictions.shape());
```
For Chinese mainland users, a mandarin video tutorial is provided on Bilibili, follow [@SomeB1oody](https://space.bilibili.com/1349872478)
对于中国大陆用户，B站有视频教程，关注[@SomeB1oody](https://space.bilibili.com/1349872478)
## Feature Flags | 特性标志
The crate uses feature flags for modular compilation:
该crate使用特性标志进行模块化编译：
| `machine_learning` | Classical ML algorithms (depends on `math`) | 经典机器学习算法（依赖于`math`） |
| `neural_network` | Neural network framework | 神经网络框架 |
| `utility` | Data preprocessing and dimensionality reduction | 数据预处理和降维 |
| `metric` | Evaluation metrics | 评估指标 |
| `math` | Mathematical utilities | 数学工具 |
| `dataset` | Standard datasets | 标准数据集 |
| `default` | Enables `machine_learning` and `neural_network` | 开启`machine_learning`和`neural_network`功能 |
| `full` | Enables all features | 启用所有功能 |
## Project Status | 项目状态
RustyML is in active development. While the API is stabilizing, breaking changes may occur in minor version updates until version 1.0.0.
RustyML正在积极开发中。虽然API正在稳定，但在1.0.0版本之前，次要版本更新中可能会出现破坏性更改。
## Contribution | 贡献
Contributions are welcome! If you're interested in helping build a robust machine learning ecosystem in Rust, please feel free to:
欢迎贡献！如果您有兴趣帮助构建Rust中的强大机器学习生态系统，请随时：
3. Provide feedback on the API design | 提供API设计的反馈意见
4. Help with documentation and examples | 帮助完善文档和示例
## Authors | 作者
SomeB1oody – [stanyin64@gmail.com](mailto:stanyin64@gmail.com)
## License | 许可证
Licensed under the [MIT License](LICENSE). See LICENSE file for details.
根据[MIT许可证](LICENSE)授权。有关详细信息，请参阅LICENSE文件。