Unsupervised learning for feature projection: Extracting patterns from multidimensional building measurements
Energy and Buildings(2020)
摘要
•Three conventional dimensionality reduction algorithms are compared.•PCA is unlikely to distort the original dataset at the cost of information loss.•Autoencoder provides a good balance between data compression and exploitation.•t-SNE is better in representing well-separated clusters than multi-collinearity.
更多查看译文
关键词
Unsupervised learning,Building performance,Dimensionality reduction,Data representation
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要