Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
37 changes: 37 additions & 0 deletions docs/RBM.md
Original file line number Diff line number Diff line change
Expand Up @@ -42,3 +42,40 @@
- 推荐系统
- 协同过滤
- 深度信念网络的基础组件
-
## 代码示例

使用 PyTorch 实现简单的 RBM:

```python
import torch
import torch.nn as nn

class RBM(nn.Module):
"""受限玻尔兹曼机"""
def __init__(self, visible_dim, hidden_dim):
super(RBM, self).__init__()
self.W = nn.Parameter(torch.randn(hidden_dim, visible_dim) * 0.01)
self.b_v = nn.Parameter(torch.zeros(visible_dim))
self.b_h = nn.Parameter(torch.zeros(hidden_dim))

def sample_hidden(self, v):
"""从可见层采样隐藏层"""
p_h = torch.sigmoid(torch.matmul(v, self.W.t()) + self.b_h)
return p_h, torch.bernoulli(p_h)

def sample_visible(self, h):
"""从隐藏层采样可见层"""
p_v = torch.sigmoid(torch.matmul(h, self.W) + self.b_v)
return p_v, torch.bernoulli(p_v)
```

## 运行环境

- Python 3.7+
- PyTorch 1.8+

## 参考资料

- [PyTorch 官方文档](https://pytorch.org/docs/stable/index.html)
- [深度信念网络介绍](https://www.cs.toronto.edu/~hinton/absps/fastnc.pdf)
31 changes: 31 additions & 0 deletions docs/gaussian_mixture.md
Original file line number Diff line number Diff line change
Expand Up @@ -44,3 +44,34 @@
- 异常检测
- 图像分割
- 语音识别
-
## 数学公式

GMM 的概率密度函数为:

`p(x) = Σ π_k * N(x | μ_k, Σ_k)`

其中:
- K 为高斯成分数量
- π_k 为第 k 个成分的混合权重,满足 Σπ_k = 1
- N(x | μ_k, Σ_k) 为第 k 个高斯分布

## 代码示例

使用 scikit-learn 拟合高斯混合模型:

```python
from sklearn.mixture import GaussianMixture
import numpy as np

# 生成示例数据
X = np.random.randn(300, 2)

# 创建并训练模型
gmm = GaussianMixture(n_components=3, random_state=0)
gmm.fit(X)

# 预测类别
labels = gmm.predict(X)
print("各成分权重:", gmm.weights_)
```
11 changes: 7 additions & 4 deletions ignore_users.json
Original file line number Diff line number Diff line change
@@ -1,5 +1,8 @@
[
"Haidong Wang",
"donghaiwang",
"whd@hutb.edu.cn"
]
{
"name": "Haidong Wang",
"github": "donghaiwang",
"email": "whd@hutb.edu.cn",
"role": "author"
}
]