矩阵分解系列三:非负矩阵分解及Python实现

非负矩阵分解的定义及理解

摘自《迁移学习》K-Means算法&非负矩阵三因子分解(NMTF)

下图可帮助理解:

 

举个简单的人脸重构例子: 

Python实例:用非负矩阵分解提取人脸特征

「摘自Python机器学习应用

在sklearn库中,可以使用sklearn.decomposition.NMF加载NMF算法,主要参数有:
  • n_components:指定分解后基向量矩阵W的基向量个数k
  • init:W矩阵和Z矩阵的初始化方式,默认为‘nndsvdar’

目标:已知Olivetti人脸数据共400个,每个数据是64*64大小。由于NMF分解得到的W矩阵相当于从原始矩阵中提取的特征,那么就可以使用NMF对400个人脸数据进行特征提取。
程序如下:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
from numpy.random import RandomState
import matplotlib.pyplot as plt
from sklearn.datasets import fetch_olivetti_faces   #加载Olivetti人脸数据集导入函数
from sklearn import decomposition
 
n_row, n_col = 2, 3
n_components = n_row * n_col   #设置提取的特征的数目
image_shape = (64, 64)   #设置展示时人脸数据图片的大小
 
dataset = fetch_olivetti_faces(shuffle=True, random_state=RandomState(0))
faces = dataset.data
 
def plot_gallery(title, images, n_col=n_col, n_row=n_row):
    plt.figure(figsize=(2. * n_col, 2.26 * n_row))
    plt.suptitle(title, size=16)
 
    for i, comp in enumerate(images):
        plt.subplot(n_row, n_col, i + 1)
        vmax = max(comp.max(), -comp.min())
        plt.imshow(comp.reshape(image_shape), cmap=plt.cm.gray,
                   interpolation='nearest', vmin=-vmax, vmax=vmax)
        plt.xticks(())
        plt.yticks(())
    plt.subplots_adjust(0.01, 0.05, 0.99, 0.94, 0.04, 0.)
 
plot_gallery("First centered Olivetti faces", faces[:n_components])
 
estimators = [('Eigenfaces - PCA using randomized SVD', decomposition.PCA(n_components=n_components, whiten=True)),
              ('Non-negative components - NMF', decomposition.NMF(n_components=n_components, init='nndsvda', tol=5e-3))]
 
for name, estimator in estimators:
    print("Extracting the top %d %s..." % (n_components, name))
    print(faces.shape)
    estimator.fit(faces)   #调用PCA或NMF提取特征
    components_ = estimator.components_   #获取提取的特征
    plot_gallery(name, components_[:n_components])
 
plt.show()  
运行结果:

Extracting the top 6 Eigenfaces - PCA using randomized SVD...
(400, 4096)
Extracting the top 6 Non-negative components - NMF...
(400, 4096)

   
 
posted @   Picassooo  阅读(2350)  评论(0编辑  收藏  举报
编辑推荐:
· 基于Microsoft.Extensions.AI核心库实现RAG应用
· Linux系列:如何用heaptrack跟踪.NET程序的非托管内存泄露
· 开发者必知的日志记录最佳实践
· SQL Server 2025 AI相关能力初探
· Linux系列:如何用 C#调用 C方法造成内存泄露
阅读排行:
· Manus爆火,是硬核还是营销?
· 终于写完轮子一部分:tcp代理 了,记录一下
· 别再用vector<bool>了!Google高级工程师:这可能是STL最大的设计失误
· 震惊!C++程序真的从main开始吗?99%的程序员都答错了
· 单元测试从入门到精通
点击右上角即可分享
微信分享提示