The estimated covariance matrix of the parameter estimates is computed as the inverse Hessian matrix, and for unconstrained problems it should be positive definite. If the final parameter estimates ...

To address this, this paper proposes an improved RIFT algorithm based on pre-screening acceleration: First, a fast feature point pre-screening mechanism using Hessian matrix is designed to rapidly ...

Semi-supervised sparse feature selection, which can exploit the small number labeled data and large number unlabeled data simultaneously, has become an important technique in many applications on ...