81聚类Kmeans要点.ppt

从聚类到Unsupervised Learning Unsupervised learning Whitten (PCA) K-means clustering For Chinese Text For English-and-Chinese Text Adam Coates， Honglak Lee， Adrew. N.G “An Analysis of Single-Layer Networks in Unsupervised Feature Learning,” NIPS 2011. 从聚类到Unsupervised Learning Learning framework 1. Extract random patches from unlabeled training images. 2. Apply a pre-processing stage to the patches. 3. Learn a feature-mapping using an unsupervised learning algorithm. Feature extraction procedures 1. Extract features from equally spaced sub-patches covering the input image. 2. Pool features together over regions of the input image to reduce the number of feature values. 3. Train a linear classifier to predict the labels given the feature vectors. Adam Coates， Honglak Lee， Adrew. N.G “An Analysis of Single-Layer Networks in Unsupervised Feature Learning,” NIPS 2011. Feature Extraction with unsupervised learning Filtering Pooling 从聚类到Unsupervised Learning Adam Coates， Honglak Lee， Adrew. N.G “An Analysis of Single-Layer Networks in Unsupervised Feature Learning,” NIPS 2011. 从聚类到Unsupervised Learning unsupervised learning algorithms K-means clustering Sparse auto-encoder Sparse restricted Boltzmann machine Gaussian mixtures Adam Coates， Honglak Lee， Adrew. N.G “An Analysis of Single-Layer Networks in Unsupervised Feature Learning,” NIPS 2011. 从聚类到Unsupervised Learning Adam Coates， Honglak Lee， Adrew. N.G “An Analysis of Single-Layer Networks in Unsupervised Feature Learning,” NIPS 2011. K-means+Whitten K-means GMMs+Whitten GMMs Sparse autoencoder +Whitten Sparse autoencoder RBM+Whitten RBM 无监督单层学习在CIFAR-10图像数据集合上学习到的特征 从聚类到Unsupervised Learning Adam Coates， Honglak Lee， Adrew. N.G “An Analysis of Single-Layer Networks in Unsupervised Feature Learning,” NIPS 2011. 学习到的特征用于监督分类时的性能比较 其他聚类方法 基于密度的聚类… 基于网格的聚类… 统计推断与生成模型 混合高斯模型GMMs Latent Dirichlet Allocation (Topic Model) * 第8.1节聚类分析 中国科学院大学 叶齐祥 qxye@ucas.ac.cn 提 纲 概述 K-means聚类算法 聚类算法的距离度量 层次聚类算法 从聚类到Unsupe