In this paper we present a theoretical investigation and an extensive computational study of Exterior Point Simplex Algorithm (EPSA) initialization methods for the Assignment Problem (AP). We describe the exterior point algorithm using three different initialization methods. Effective implementations are explored for each initialization method. Then we perform an experimental evaluation on a la...

A proper initialization of the weights in a neural network is critical to its convergence. Current insights into weight initialization come primarily from linear activation functions. In this paper, I develop a theory for weight initializations with non-linear activations. First, I derive a general weight initialization strategy for any neural network using activation functions differentiable a...

nonnegative matrix factorization (nmf) is a common method in data mining that have been used in different applications as a dimension reduction, classification or clustering method. methods in alternating least square (als) approach usually used to solve this non-convex minimization problem.  at each step of als algorithms two convex least square problems should be solved, which causes high com...

Iterative refinement clustering algorithms are widely used in data mining area, but they are sensitive to the initialization. In the past decades, many modified initialization methods have been proposed to reduce the influence of initialization sensitivity problem. The essence of iterative refinement clustering algorithms is the local search method. The big numbers of the local minimum points w...

We propose a new weight initialization suited for arbitrary nonlinearities by generalizing previous weight initializations. The initialization corrects for the influence of dropout rates and an arbitrary nonlinearity’s influence on variance through simple corrective scalars. Consequently, this initialization does not require computing mini-batch statistics nor weight pre-initialization. This si...

European Centre for Medium-Range Weather Forecasts The need to control gravity waves in data assimilation and the general concept of initialization are discussed. The basic method of non-linear normal-mode initialization is introduced, and aspects of its implementation in models are presented, covering the vertical and horizontal modal decomposition, implicit initialization and diabatic initial...