Fisher Discriminant Analysis(FDA) is a widely used dimensionality reduction technique for fault diagnosis in industry process. Kernel FDA(KFDA) is developed to solve nonlinear problem. Small sample size(3S) problem is arisen in both FDA and KFDA. Regularized FDA is an effective solution for 3S problem. A novel kernel form of regularized FDA which transforms it to equation solving problem and expresses in the dual form is obtained and implementation procedure for fault diagnosis is given. Experimental results on the Tennessee Eastman(TE) process show that the algorithm is valid, and regularized KFDA outperforms FDA for 3S problem.
According to the structural characteristics of ontology model, this paper analyzes the concept similarity computation based on the width, depth and density among the concepts of ontology model. It considers these aspects as structural fact and combines other facts such as semantic contact ratio and semantic distance. An algorithm of semantic similarity computation based on ontology structure is proposed. Through constructing ontology model and analyzing experiment, it summarizes the effects to the ontology concepts semantic similarity of different facts about ontology structure.
Aiming at solving the problem that the dotfilter can not detect Ground Glass Opacity(GGO) nodules, a method based on an adaptive nonlinear filter and dot filter is used to detect them. Experiments are performed at 23 scans of CT images which contain 32 GGO nodule. The sensitivity of the detection method is 84.4%. The time cost is 1.2 minutes per scan. This method is superior to existing methods on the required time and sensitivity.
This paper studies the characteristics of remote sensing data and analyzes application status, combines with LOD ideas and images pyramid model and uses the optimize fragmentary image division algorithm to divide images into several blocks, according to the way of shared directory to organize the fragments. It establishes a Webbased distributed remote sensing image data release framework, achieves the rapid organization and realtime release of remote sensing images, and provides good ideas to solve the problems of sharing remote sensing image data.
This paper gives the definition of largescale complex dataset with characteristics of large, multiattribute, temporal and spatial, rough. For the problem of largescale complex dataset mining, according to theory of data reduction of statistics, rough set, fuzzy set, an efficient method is proposed to reduce largescale complex data based on fuzzy clustering and attribute reduction of rough set.
The present Parallel Variable Transformation(PVT) algorithm is not suitable for large scale optimization problem with bounded constraints. This paper proposes a parallel feasible algorithm for bounded constrained optimization problem by the interiorpoint and the gradual reduction. Yet, the convergence of the algorithm is obtained under the conditions that the reduction direction is gradient dependent and the search step satisfies Goldstein criteria. When the constraints losing effectiveness, it degenerates into the unconstrained PVT algorithm. It is an extension of unconstrained PVT algorithm.
Under the network environment for the application of virtual interaction, this paper presents a Virtual Interative Engine Model(VIEM) structural model, using multiAgent collaborative and mutual inductance 3D interactive interface standard for virtual reality system. The Agent database, interaction rule base, etc. and 4 core module structure database are provided. Through interactive database, virtual interactive subsystems are built, such as user input operation of perception, and a virtual interactive engine solution is given.