Coupled Tensor for Data Analysis

Tensor component analysis, which can reveal the underlying structure of multiway data and exploit the relationship among multiple modes, plays an important role in signal processing and machine learning. However, with the development of sensors, we can obtain data that share common latent information but also maintain certain independent structures. For such data, traditional tensor-based methods process them in an independent manner, resulting in poor performance especially when some entries of data are missing.