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。

Overview | 概述

Rust AI 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 with foundational components, the project's long-term vision is to provide a complete ecosystem for machine learning, deep learning, and transformer-based models. Rust AI 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，只实现了基础组件，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。

Current Features | 当前功能

• Mathematical Utilities | 数学工具:

  • Sum of square total (SST) for measuring data variability | 总平方和(SST)，用于衡量数据变异性
  • Sum of squared errors (SSE) for evaluating prediction errors | 误差平方和(SSE)，用于评估预测误差
  • Sigmoid function for logistic regression and neural networks | Sigmoid函数，用于逻辑回归和神经网络
  • Logistic loss (log loss) for binary classification models | 逻辑损失函数(对数损失)，用于二元分类模型
  • Accuracy score for classification model evaluation | 准确率分数，用于分类模型评估
  • Calculate the squared Euclidean distance between two points | 计算两点之间的欧几里得距离平方
  • Calculate the Manhattan distance between two points ｜ 计算两点间的曼哈顿距离
  • Calculate the Minkowski distance between two points ｜ 计算两点间的闵可夫斯基距离
  • Calculate the variance | 计算方差
  • Calculates the entropy of a label set | 计算标签集的熵
  • Calculates the Gini impurity of a label set | 计算标签集的基尼不纯度
  • Calculates the information gain when splitting a dataset | 计算数据集分割时的信息增益
  • Calculates the gain ratio for a dataset split | 计算数据集分割的增益率
  • Calculates the leaf node adjustment factor c(n) | 计算叶节点调整因子 c(n)
  • Calculates the standard deviation of a set of values | 计算标准差

• Metric Utilities | 度量工具:

  • R-squared (R²) score for assessing model fit quality | R平方(R²)分数，用于评估模型拟合质量
  • Calculates the Root Mean Squared Error (RMSE) between predicted and actual values | 计算均方误差根
  • Calculates the Mean Squared Error (MSE) of a set of values | 计算一组值的均方误差(MSE)
  • Calculates the Mean Absolute Error (MAE) between predicted and actual values | 计算平均绝对误差
  • Confusion Matrix for binary classification evaluation | 混淆矩阵及相关计算公式
  • Calculates the Normalized Mutual Information (NMI) between two cluster label assignments.(Unit: nat) | 计算NMI值(单位: nat)
  • Calculates the Adjusted Mutual Information (AMI) between two cluster label assignments.(Unit: nat) | 计算AMI值(单位: nat)
  • Calculates the Area Under the Receiver Operating Characteristic Curve (AUC-ROC) | 计算AUC-ROC值
  • Calculates the Mean Squared Error (MSE) of a set of values | 计算一组值的均方误差(MSE)

• Machine Learning Models | 机器学习模型:

  • Supervised Learning | 监督式学习:

    • Linear Regression | 线性回归
    • Logistic Regression | 逻辑回归
    • KNN | K邻近值聚类
    • Decision Tree | 决策树
    • SVC(Support Vector Classification) | 支持向量机分类
    • Linear SVC | 线性向量机分类
    • LDA(Linear Discriminant Analysis) | 线性判别分析
    • Kernel PCA | 核主成分分析

  • Unsupervised Learning | 无监督学习:

    • KMeans | K均值聚类
    • MeanShift | MeanShift聚类
    • DBSCAN | DBSCAN聚类
    • Isolation Forest | 隔离森林

• Utilities | 工具:

  • PCA(Principal Component Analysis) | 主成分分析
  • standardize data($\mu = 0$, $\sigma = 1$) | 标准化数据(确保均值$\mu = 0$, 协方差$\sigma = 1$)
  • Split dataset for training and dataset for test | 分离训练集和测试集
  • LDA(Linear Discriminant Analysis) | 线性判别分析
  • t-Distributed Stochastic Neighbor Embedding (t-SNE) | t分布随机邻域嵌入

• Datasets | 数据集:

  • Iris dataset | 鸢尾花分类数据集
  • Diabetes dataset | 糖尿病回归数据集

• Neural Network(initial implementation) | 神经网络(初步实现):

  • model | 模型:

    • Sequential | 顺序模型

  • Layer | 层:

    • Dense | 全连接层

  • Loss function | 损失计算函数:

    • Mean Squared Error | 均方误差
    • 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

  • Optimizer | 优化器:

    • SGD | 随机梯度下降
    • Adam | 自适应矩估计
    • RMSprop | 均方根传播

  • Activation Function | 激活函数:

    • ReLU | 修正线性单元（线性整流函数）
    • Tanh | 双曲正切函数
    • Sigmoid | 逻辑函数
    • Softmax | 归一化指数函数

Vision | 愿景

While the library is in its early stages, Rust AI aims to evolve into a comprehensive crate that includes: 虽然该库处于早期阶段，但Rust AI旨在发展成为一个包含以下内容的综合性crate：

• Classical Machine Learning Algorithms | 经典机器学习算法:

  • Linear and Logistic Regression | 线性和逻辑回归
  • Decision Trees | 决策树
  • Clustering algorithms (K-means, MeanShift, KNN) | 聚类算法(K均值, MeanShift, KNN)
  • Dimensionality reduction technique (PCA) | 降维技术(PCA)

• Deep Learning | 深度学习:

  • Neural network building blocks | 神经网络构建模块
  • Convolutional neural networks | 卷积神经网络
  • Recurrent neural networks | 循环神经网络
  • Optimization algorithms | 优化算法

• Transformer Architecture | Transformer架构:

  • Self-attention mechanisms | 自注意力机制
  • Multi-head attention | 多头注意力
  • Encoder-decoder architectures | 编码器-解码器架构
  • Pre-training and fine-tuning capabilities | 预训练和微调能力

• Utilities | 实用工具:

  • Data preprocessing | 数据预处理
  • Cross-validation | 交叉验证
  • Performance metrics | 性能指标
  • Visualization helpers | 可视化辅助工具

Dependencies | 依赖

• ndarray (0.16.1): N-dimensional array library for Rust | Rust的N维数组库
• rand (0.9.0): Random number generators and other randomness functionality for Rust | Rust的随机数生成器和其他随机性功能
• nalgebra (0.33.2): General-purpose linear algebra library with transformations and statically-sized or dynamically-sized matrices | 通用线性代数库，具有变换和静态大小或动态大小的矩阵
• statrs(0.18.0): A host of statistical utilities for Rust scientific computing | Rust 科学计算的一整套统计工具
• rand_distr(0.5.1) Sampling from random number distributions | 从随机数分布中采样
• rayon(1.10.0) a data-parallelism library for Rust | Rust的数据并行库

Getting Started | 开始使用

Add the library to your Cargo.toml: 将库添加到您的Cargo.toml文件中：

[dependencies]
rustyml = "0.5.0"

Example usage | 使用示例:

use rustyml::machine_learning::linear_regression::LinearRegression;
use ndarray::{Array1, Array2, array};

// Create a linear regression model
let mut model = LinearRegression::new(true, 0.01, 1000, 1e-6);

// 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());

// 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);

For Chinese mainland users, I provide mandarin video tutorial on Bilibili, follow @SomeB1oody 对于中国大陆用户,我在B站提供视频教程,关注@SomeB1oody

Project Status | 项目状态

This project is in the early development stage. Currently, only a small subset of the planned functionality has been implemented. The API is unstable and subject to significant changes as the project evolves. 该项目处于早期开发阶段。目前，仅实现了计划功能的一小部分。API不稳定，随着项目的发展可能会有重大变化。

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：

1. Submit issues for bugs or feature requests | 提交bug或功能请求
2. Create pull requests for improvements | 创建改进的拉取请求
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

License | 许可证

MIT - License

Authors | 作者

• SomeB1oody (stanyin64@gmail.com)

RustyML - Bringing the power and safety of Rust to machine learning and AI. RustyML - 将Rust的强大性能和安全特性引入机器学习和人工智能领域。