proximal variable metric method with spectral gradient updated for large scale sparse optimizationA new method called spectral proximal method (SPM) is proposed, which incorporates the Diagonal variable metric method with the proximal method to solve large-scale sparse optimization problems. By considering the trade-off between efficiencies and the sparsity of the solutions, SPM method is more applicable compared to other proximal gradient methods to large-scale real-world datasets. This method is useful in the stock portfolio optimization and machine learning areas.
Date of Conceived : jun 25, 2023
Date of Declared : aug 28, 2023