The Hanson-Wright inequality for random tensors (Q2164658): Difference between revisions

The Hanson-Wright inequality is an important probabilistic tool and can be found in various textbooks covering the basics of signal processing and probability theory. It has found numerous applications, in particular it has been a key ingredient for the construction of fast Johnson-Lindenstrauss embeddings. The main results of the present paper provide moment estimates for the semi-decoupled chaos process that is the sum of the exspressions of the type \[ (X(1)\otimes\cdots\otimes X(d))^T A(X(1)\otimes\cdots\otimes X(d)), \] where \(\otimes\) denotes the Kronecker product and \(X(1), \dots , X(d)\) are random vectors with independent, zero mean subgaussian entries with unit variance. In other words, the Hanson-Wright inequality for random tensors is obtained. The bounds are tight up to constants depending on \(d\) for the case of Gaussian random vectors. The proof also provides a decoupling inequality for expressions of the mentioned type. Using these bounds, the authors obtain new, improved concentration inequalities. The results also lay the foundations for an order-optimal analysis of fast Kronecker-structured Johnson-Lindenstrauss embeddings.