# Paper_projects

**Repository Path**: lis-kust/paper_projects

## Basic Information

- **Project Name**: Paper_projects
- **Description**: 一些经典论文的复现或代码整理
- **Primary Language**: Unknown
- **License**: MulanPSL-2.0
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2026-01-07
- **Last Updated**: 2026-04-09

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Paper_projects
### 一些论文算法的复现和整理, 包括但不限于：pdf原文， 代码实现， 样例数据等等。

## 1, Echo State Network







## 2, Stochastic Configuration Network







## 3, Granular Ball Sets



### 3.1,  基于粒球的三支聚类 (Granular Ball Three-Way Clustering)

该实现结合了传统的K-means聚类和三支决策思想。它通过将数据点聚合成“粒球”来形成信息粒，然后根据粒球的纯度及其与近邻粒球的关系，将样本划分为核心域（Positive Region）、边界域（Boundary Region）和琐碎域（Negative Region）。这有助于在面对不确定性信息时做出更精细的决策。

This implementation combines traditional K-means clustering with the concept of three-way decisions. It forms information granules by aggregating data points into "granular balls." Then, based on the purity of these granular balls and their relationships with neighboring granular balls, samples are partitioned into Positive Region, Boundary Region, and Negative Region. This approach facilitates more refined decision-making when dealing with uncertain information.

**文件**: `GranularThreeWayClustering.py` 

**核心特点**:

- **粒球建模**: 每个簇被抽象为一个具有中心和半径的粒球。
- **区域划分**: 根据粒球纯度，将样本划分为核心域、边界域和琐碎域。
- **简化决策**: 在不确定性下提供更灵活的分类。

**File**: `GranularThreeWayClustering.py`

**Key Features**:

- **Granular Ball Modeling**: Each cluster is abstracted as a granular ball with a center and a radius.
- **Region Partitioning**: Samples are divided into Positive, Boundary, and Negative regions based on granular ball purity.
- **Refined Decision-Making**: Provides more flexible classification under uncertainty.



---

### 3.2, 基于粒球邻域粗糙集的三支高斯混合聚类 (Granular Ball Neighborhood Rough Set Gaussian Mixture Three-Way Clustering)

此算法将高斯混合模型（GMM）的概率聚类能力与邻域粗糙集（NRS）理论和三支决策相结合。它利用GMM进行初始聚类和概率估计，然后通过定义数据点的邻域来构建粗糙集（下近似和上近似），最终根据GMM的后验概率和粗糙集近似结果进行三支决策划分。

This algorithm integrates the probabilistic clustering capabilities of Gaussian Mixture Models (GMM) with Neighborhood Rough Set (NRS) theory and three-way decision. It uses GMM for initial clustering and probability estimation, then constructs rough sets (lower and upper approximations) by defining neighborhoods for data points, ultimately performing three-way decision partitioning based on GMM posterior probabilities and rough set approximations.

**文件**: `GBNRSGMMThreeWayClustering.py`

**核心特点**:

- **GMM驱动**: 使用高斯混合模型进行软聚类和概率输出。
- **邻域粗糙集**: 基于数据点邻域定义粗糙集上下近似，捕捉数据结构。
- **概率三支决策**: 结合GMM概率和粗糙集近似，进行核心域、边界域、琐碎域划分。

**File**: `GBNRSGMMThreeWayClustering.py`

**Key Features**:

- **GMM-Driven**: Utilizes Gaussian Mixture Models for soft clustering and probabilistic output.
- **Neighborhood Rough Set**: Defines lower and upper approximations based on data point neighborhoods to capture data structure.
- **Probabilistic Three-Way Decision**: Combines GMM probabilities and rough set approximations for partitioning into Positive, Boundary, and Negative regions.



---

### 3.3, 多粒度三支决策 (Multi-Granular Three-Way Decision)

该方法通过在多个“粒度”下执行三支聚类，并融合这些粒度下的决策结果，以实现更鲁棒和全面的聚类划分。每个“粒度”可以由不同的邻域半径（`epsilon`）和/或不同的特征子集（`features`）来定义，从而从多个视角分析数据的不确定性。

This method performs three-way clustering across multiple "granularities" and then fuses the decision results from these different granularities to achieve a more robust and comprehensive clustering partition. Each "granularity" can be defined by a different neighborhood radius (`epsilon`) and/or a different subset of features (`features`), allowing for the analysis of data uncertainty from multiple perspectives.

**文件**: `MultiGranularThreeWayClustering.py`

**核心特点**:

- **多视图分析**: 支持在不同邻域尺度和不同特征子集上进行聚类。
- **决策融合**: 整合来自多个粒度的决策结果，提高最终划分的稳健性。
- **综合不确定性处理**: 更全面地捕捉数据中的模糊和不确定信息。

**File**: `MultiGranularThreeWayClustering.py`

**Key Features**:

- **Multi-View Analysis**: Supports clustering at different neighborhood scales and with different feature subsets.
- **Decision Fusion**: Integrates decision results from multiple granularities to enhance the robustness of the final partitioning.
- **Comprehensive Uncertainty Handling**: Captures ambiguity and uncertainty in data more thoroughly.

### 如何运行代码 (How to Run the Code)

1. **环境准备 (Environment Setup)**:

   - 确保您已安装Python 3.x。

   - 安装所需的库: `numpy`, `scikit-learn`, `matplotlib`。

     ```
     pip install -r requirments.txt
     ```

2. **运行示例 (Running Examples)**:

   - 每个Python文件都包含一个 `if __name__ == "__main__":` 块，其中提供了使用该算法的示例代码。

   - 打开您选择的文件，例如 `MultiGranularThreeWayClustering.py`。

   - 在终端中运行该文件:

     ```
     python MultiGranularThreeWayClustering.py
     ```

   - 程序将生成模拟数据，执行聚类，并在控制台打印结果摘要。如果安装了 `matplotlib`，还将显示可视化图表。













## 参考文献 (References)

以下是这些方法所基于的原始文献引用：

Here are the original literature citations that these methods are based on:









3.1, **基于粒球的三支聚类**:

- 韩兴雨, 朱金, 孟义平, 王平心. 基于粒球的三支聚类方法. 江苏科技大学学报(自然科学版), 2024, 38(5): 57-62.
- Xia, S., Liu, Y., Ding, X., Wang, G., Yu, H., & Luo, Y. (2019). Granular ball computing classifiers for efficient, scalable and robust learning. *Information Sciences*, 483, 136-152.

3.2, **基于粒球邻域粗糙集的三支高斯混合聚类**:

- 邵春梅, 万仁霞, 苗夺谦, 赵杰. 基于粒球邻域粗糙集的三支高斯混合聚类. 郑州大学学报(理学版). DOI: 10.13705/j.issn.1671-6841.2024108.
- 万仁霞, 王大庆, 苗夺谦. 基于三支决策的高斯混合聚类研究. 重庆邮电大学学报(自然科学版), 2021, 33(5): 806-815.

3.3, **多粒度三支决策**:

- Chen, J., Zhang, Y. P., & Zhao, S. (2016). Multi-granular mining for boundary regions in three-way decision theory. *Knowledge-Based Systems*, 91, 287-292.
- Yao, Y. (2013). *Granular Computing and Sequential Three-Way Decisions*. Springer Berlin Heidelberg.
- 陈中华, 巴婧, 徐泰华, 王平心, 杨习贝. 一种面向粒球粗糙集的快速约简求解方法. 小型微型计算机系统, 2023, 44(1): 24-29.