Tensor Decompositions via Two-Mode Higher-Order SVD (HOSVD)
This work addresses a specific problem in tensor decomposition for statistics and machine learning, offering incremental improvements in noise handling and accuracy.
The paper tackles the problem of decomposing symmetric, nearly orthogonally decomposable tensors by introducing a new method based on two-mode unfoldings and rank-1 constraints, which provably handles more noise and achieves high estimation accuracy, with numerical results showing robustness and improved performance as tensor order increases.
Tensor decompositions have rich applications in statistics and machine learning, and developing efficient, accurate algorithms for the problem has received much attention recently. Here, we present a new method built on Kruskal's uniqueness theorem to decompose symmetric, nearly orthogonally decomposable tensors. Unlike the classical higher-order singular value decomposition which unfolds a tensor along a single mode, we consider unfoldings along two modes and use rank-1 constraints to characterize the underlying components. This tensor decomposition method provably handles a greater level of noise compared to previous methods and achieves a high estimation accuracy. Numerical results demonstrate that our algorithm is robust to various noise distributions and that it performs especially favorably as the order increases.