TensorComplete

Efficient algorithms for tensor noise reduction and completion. This package includes a suite of parametric and nonparametric tools for estimating tensor signals from noisy, possibly incomplete observations. The methods allow a broad range of data types, including continuous, binary, and ordinal-valued tensor entries. The algorithms employ the alternating optimization. The detailed algorithm description can be found in the following three references.