Self-stabilized minor subspace extraction algorithm based on Householder transformation

In this paper, we propose an orthogonalized version of Oja's algorithm (OOja) that can be used for the estimation of minor and principal subspaces of a vector sequence. The new algorithm offers, as compared to Oja, such advantages as orthogonality of the weight matrix, which is ensured at each iteration, numerical stability and a quite similar computational complexity.