Nonnegative Matrix Factorization (NMF) is a widely used technique in many applications such as face recognition, motion segmentation, etc. It approximates the nonnegative data in an original high dimensional space wit...
详细信息
ISBN:
(纸本)9781479957521
Nonnegative Matrix Factorization (NMF) is a widely used technique in many applications such as face recognition, motion segmentation, etc. It approximates the nonnegative data in an original high dimensional space with a linear representation in a low dimensional space by using the product of two nonnegative matrices. In many applications data are often partially corrupted with large additive noise. When the positions of noise are known, some existing variants of NMF can be applied by treating these corrupted entries as missing values. However, the positions are often unknown in many real world applications, which prevents the usage of traditional NMF or other existing variants of NMF. This paper proposes a Robust Nonnegative Matrix Factorization (RobustNMF) algorithm that explicitly models the partial corruption as large additive noise without requiring the information of positions of noise. In particular, the proposed method jointly approximates the clean data matrix with the product of two nonnegative matrices and estimates the positions and values of outliers/noise. An efficient iterative optimization algorithm with a solid theoretical justification has been proposed to learn the desired matrix factorization. Experimental results demonstrate the advantages of the proposed algorithm.
This paper studied the Vehicle Scheduling Problem under the managerial thinking of Vendor Managed Inventory (abbreviated as VMI). It particularly established the mathematical model and the iterativeoptimization algor...
详细信息
This paper studied the Vehicle Scheduling Problem under the managerial thinking of Vendor Managed Inventory (abbreviated as VMI). It particularly established the mathematical model and the iterative optimization algorithm to deal with the issues of integration of inventory management and distribution transport under VMI mode, and proved that these approaches can play roles in optimization of reducing distribution cost for logistics distribution system of supply chain.
An iterative optimization algorithm for designing two dimensions, finite aperture, aperiodic diffractive micro-optical elements based on rigorous electromagnetic computation model-the finite-difference time-domain met...
详细信息
An iterative optimization algorithm for designing two dimensions, finite aperture, aperiodic diffractive micro-optical elements based on rigorous electromagnetic computation model-the finite-difference time-domain method has been proposed. All aspects relating to the algorithm such as the finite-difference time-domain method, and the optimization process have been discussed in detail. Without any approximation based on scalar theory, the algorithm can present rigorous design results with reasonable computational cost. An aspherical surface lens and four kinds of off-axis lenses for normally incident TE mode have been designed to illustrate our algorithm. Meanwhile, we also consider the influence of quantized-lenses profiles on their diffraction performance. (C) 2003 Elsevier Science B.V. All rights reserved.
An efficient electromagnetic method for the optimization of shaped reflectors is presented. The repeated computation of scattering from reflector surface at each iteration, using the conventional numerical techniques ...
详细信息
An efficient electromagnetic method for the optimization of shaped reflectors is presented. The repeated computation of scattering from reflector surface at each iteration, using the conventional numerical techniques of either physical optics (PO) or aperture integration (AI), usually makes iterative procedures very inefficient for the synthesis of large reflectors. Recently, an asymptotic Gaussian beam (GB) technique was developed and applied successfully to the fast analysis of reflector antennas of various shapes. This GB technique completely avoids numerical integration and thus makes the analysis very efficient. Our method uses the GB technique, coupled with a local description of the primary source and the reflector. The primary source radiation is expanded using a modified Gaussian beams basis. An antenna pattern calculation is demonstrated on a reflector that is described by local parameters in a novel way. By virtue of these local properties and the use of a steep step descent algorithm, a basic display of antenna pattern optimization is presented to illustrate the effectiveness of our method.
The inverse of the parameters of underground targets is still highly in expectation in Ground Penetrating Radar(GPR) applications. In this paper, the FDTD method and time-domain optimizationalgorithm have been used t...
详细信息
ISBN:
(纸本)0780370317
The inverse of the parameters of underground targets is still highly in expectation in Ground Penetrating Radar(GPR) applications. In this paper, the FDTD method and time-domain optimizationalgorithm have been used to reconstruct the permittivities of underground multiple targets. An iterative optimization algorithm has been put forward. The forward scattering data are also calculated by 2.5D-FDTD algorithm in order to better simulate the practical GPR situation. The relationship between the convergent rates of the inverse algorithm and the initialization values are studied.
In recent years Image Fractal Compression techniques (IFS) have gained more interest because of their capability to achieve high compression ratios while maintaining very good quality of the reconstructed image. The m...
详细信息
In recent years Image Fractal Compression techniques (IFS) have gained more interest because of their capability to achieve high compression ratios while maintaining very good quality of the reconstructed image. The main drawback of such techniques is the very high computing time needed to determine the compressed code. In this work, after a brief description of IFS theory, we introduce the coefficient quantization problem, presenting two algorithms for its solution: the first one is based on Simulated Annealing while the second refers to a fast iterativealgorithm. We discuss IFS parallel implementation at different level of granularity and we show that Massively Parallel Processing on SIMD machines is the best way to use all the large granularity parallelism offered by the problem. The results we present are achieved implementing the proposed algorithms for IFS compression and coefficient quantization on the MPP APE100/Quadrics machine. (C) 1999 Elsevier Science B.V. All rights reserved.
暂无评论