Rank determination issue is one of the most significant issues in non-negative matrix factorization (NMF) research. However, rank determination problem has not received so much emphasis as sparseness regularization pr...Rank determination issue is one of the most significant issues in non-negative matrix factorization (NMF) research. However, rank determination problem has not received so much emphasis as sparseness regularization problem. Usually, the rank of base matrix needs to be assumed. In this paper, we propose an unsupervised multi-level non-negative matrix factorization model to extract the hidden data structure and seek the rank of base matrix. From machine learning point of view, the learning result depends on its prior knowledge. In our unsupervised multi-level model, we construct a three-level data structure for non-negative matrix factorization algorithm. Such a construction could apply more prior knowledge to the algorithm and obtain a better approximation of real data structure. The final bases selection is achieved through L2-norm optimization. We implement our experiment via binary datasets. The results demonstrate that our approach is able to retrieve the hidden structure of data, thus determine the correct rank of base matrix.展开更多
AIM： To explore a feasible method on the establishment of an animal model of conjunctivochalasis （CCH）. METHODS： Twelve clean-grade New Zealand white rabbits were divided into four groups （n=3/group）： the contr...AIM： To explore a feasible method on the establishment of an animal model of conjunctivochalasis （CCH）. METHODS： Twelve clean-grade New Zealand white rabbits were divided into four groups （n=3/group）： the control group （one received no interventions, and the others underwent subconjunctival injection of sterile water）, the matrix metalloproteinases （MMPs） group （administered subconjunctival injection of MMP-3）, the aging group （administered subcutaneous injection of D-galactose）, the tumor necrosis factor-α （TNF-α） solution group （administered eye drops of TNF-α）. Anterior segment photography, conjunctival tissue light microscopy and transmission electron microscopy （TEM） were performed after 12wk. RESULTS： Among all groups, the MMPs group had the following changes： the looser connection between the inferior bulbar conjunctiva and sclera; the more disordered collagen fibers （Trichrome staining） and the broken elastic fibers （Aldehyde-fuchsin staining）; the focal necrosis of fibroblasts （TEM）. CONCLUSION： Administration of MMPs may be a feasible method for the establishment of an animal model of CCH.展开更多
The germinal matrix being an accumulation of immature blood vessels in the premature infant brain is known to be the main cause of the intracranial hemorrhage. To investigate the injuring mechanism to the blood vessel...The germinal matrix being an accumulation of immature blood vessels in the premature infant brain is known to be the main cause of the intracranial hemorrhage. To investigate the injuring mechanism to the blood vessels of the germinal matrix, a modeling scenario that consists of three basic steps is proposed. First, the cerebral blood flow that depends on autoregulation, CO2 reactivity, and variations of intracranial pressure is modeled. Second, the chaotic blood vessel network of the germinal matrix is generated, and blood pressures in the vessels of this network are computed dependent on the outcome of the first step. In the third step, the pressures computed on the second step are used in finite element simulations of separate blood vessels of the germinal matrix to detect critical values for vessels impairment.展开更多
The characterization of reinforcement in 15% SiC particles reinforced AI matrix composites processed by powder metallurgy route was studied by statistical method. During the analysis, a new approach for the estimation...The characterization of reinforcement in 15% SiC particles reinforced AI matrix composites processed by powder metallurgy route was studied by statistical method. During the analysis, a new approach for the estimation of the characterization of reinforcement was presented. The mathematic software MATLAB was used to calculate the area and perimeter of reinforcement, in which the image processing technique was applied. Based on the calculation, the fractal dimension, shape factor, reinforcement size distribution and reinforcement distribution were investigated. The results show that the reinforcement shape is similar to rectangle; the reinforcement size distribution is broad with the＇ range of 1-12 μm; the topography of reinforcement is smooth; and the reinforcement distribution is inhomogeneous. Furthermore, the cell model based on the statistical characterization was established and tested.展开更多
文摘Rank determination issue is one of the most significant issues in non-negative matrix factorization (NMF) research. However, rank determination problem has not received so much emphasis as sparseness regularization problem. Usually, the rank of base matrix needs to be assumed. In this paper, we propose an unsupervised multi-level non-negative matrix factorization model to extract the hidden data structure and seek the rank of base matrix. From machine learning point of view, the learning result depends on its prior knowledge. In our unsupervised multi-level model, we construct a three-level data structure for non-negative matrix factorization algorithm. Such a construction could apply more prior knowledge to the algorithm and obtain a better approximation of real data structure. The final bases selection is achieved through L2-norm optimization. We implement our experiment via binary datasets. The results demonstrate that our approach is able to retrieve the hidden structure of data, thus determine the correct rank of base matrix.
基金Supported by the Key Medical Discipline Project of Shanghai Municipal Health Bureau-Ophthalmology(No.ZK2015A20)the Health System Independent Innovation Science Foundation of Shanghai Putuo District(No.2015PTKW001)Plateau Science,Integrated Traditional Chinese and Western Medicine,Shanghai University of Traditional Chinese Medicine
文摘AIM： To explore a feasible method on the establishment of an animal model of conjunctivochalasis （CCH）. METHODS： Twelve clean-grade New Zealand white rabbits were divided into four groups （n=3/group）： the control group （one received no interventions, and the others underwent subconjunctival injection of sterile water）, the matrix metalloproteinases （MMPs） group （administered subconjunctival injection of MMP-3）, the aging group （administered subcutaneous injection of D-galactose）, the tumor necrosis factor-α （TNF-α） solution group （administered eye drops of TNF-α）. Anterior segment photography, conjunctival tissue light microscopy and transmission electron microscopy （TEM） were performed after 12wk. RESULTS： Among all groups, the MMPs group had the following changes： the looser connection between the inferior bulbar conjunctiva and sclera; the more disordered collagen fibers （Trichrome staining） and the broken elastic fibers （Aldehyde-fuchsin staining）; the focal necrosis of fibroblasts （TEM）. CONCLUSION： Administration of MMPs may be a feasible method for the establishment of an animal model of CCH.
文摘The germinal matrix being an accumulation of immature blood vessels in the premature infant brain is known to be the main cause of the intracranial hemorrhage. To investigate the injuring mechanism to the blood vessels of the germinal matrix, a modeling scenario that consists of three basic steps is proposed. First, the cerebral blood flow that depends on autoregulation, CO2 reactivity, and variations of intracranial pressure is modeled. Second, the chaotic blood vessel network of the germinal matrix is generated, and blood pressures in the vessels of this network are computed dependent on the outcome of the first step. In the third step, the pressures computed on the second step are used in finite element simulations of separate blood vessels of the germinal matrix to detect critical values for vessels impairment.
文摘The characterization of reinforcement in 15% SiC particles reinforced AI matrix composites processed by powder metallurgy route was studied by statistical method. During the analysis, a new approach for the estimation of the characterization of reinforcement was presented. The mathematic software MATLAB was used to calculate the area and perimeter of reinforcement, in which the image processing technique was applied. Based on the calculation, the fractal dimension, shape factor, reinforcement size distribution and reinforcement distribution were investigated. The results show that the reinforcement shape is similar to rectangle; the reinforcement size distribution is broad with the＇ range of 1-12 μm; the topography of reinforcement is smooth; and the reinforcement distribution is inhomogeneous. Furthermore, the cell model based on the statistical characterization was established and tested.
