作者: T. Heskes
DOI: 10.1109/72.963766
关键词: Self-organizing map 、 Data mining 、 Artificial intelligence 、 Pattern recognition 、 Visualization 、 Vector quantization 、 Unsupervised learning 、 Data visualization 、 Multinomial distribution 、 Computer science 、 Missing data
摘要: Self-organizing maps are popular algorithms for unsupervised learning and data visualization. Exploiting the link between vector quantization mixture modeling, we derive expectation-maximization (EM) self-organizing with without missing values. We compare elastic-net approach explain why former is better suited visualization of high-dimensional data. Several extensions improvements discussed. As an illustration apply a map based on multinomial distribution to market basket analysis.