Dianxun Shuai

A new data clustering approach: Generalized cellular automata

This paper is devoted to a novel stochastic generalized cellular automata (GCA) for self-organizing data clustering in enterprise computing. The GCA transforms the data clustering process into a stochastic process over the configuration space on a GCA array. The GCA-based approach to data clustering has many advantages in terms of the real-time performance and the ability to effectively deal with a variety of data sets, including noise data, dynamical data, multi-type and multi-distribution data, high-dimensional and large-scale data set. The GCA clustering approach also has the learning ability, and the better feasibility for hardware implementation with VLSI systolic technology. The simulations and comparisons have shown the effectiveness of the proposed GCA for the data clustering in enterprise computing.

Particle Model to Optimize Enterprise Computing

This paper presents a novel generalized particle model (GPM) for the parallel optimization of enterprise computing. Since enterprise computing always involves the resource allocation, task assignment, and behavior coordination, without loss of generality, the proposed GPM is devoted to the optimization of enterprise computing in the context of the resource allocation and task assignment in complex environment. GPM transforms the optimization of enterprise computing into the kinematics and dynamics of massive particles in a force-field. The GPM approach has many advantages in terms of the high-scale parallelism, multi-objective optimization, multi-type coordination, multi-degree personality, and the ability to handle complex factors. Simulations have shown the effectiveness and suitability of the proposed GPM approach to optimize the enterprise computing.

Particle model to optimize resource allocation and task assignment

This paper presents a novel generalized particle model for the parallel optimization of the resource allocation and task assignment in complex environment of enterprise computing. The generalized particle model (GPM) transforms the optimization process into the kinematics and dynamics of massive particles in a force-field. The GPM approach has many advantages in terms of the high-scale parallelism, multi-objective optimization, multi-type coordination, multi-degree personality, and the ability to handle complex factors. Simulations show the effectiveness and suitability of the proposed GPM approach to optimize the enterprise computing.

Problem-Solving in Multi-Agent Systems: A Novel Generalized Particle Model

This paper presents a novel generalized particle model (GPM) for problem-solving in multi-agent systems (MAS). The construction, dynamics and properties of the GPA and corresponding algorithm are discussed. The GPA has many advantages in terms of the high-scale parallelism, multi-objective optimization, multi-type coordination, multi-degree autonomy, and the ability to deal randomly occurring phenomena in MAS systems

Generalized Cellular Automata For Data Clustering

This paper is devoted to novel stochastic generalized cellular automata (GCA) for self-organizing data clustering. The GCA transforms the data clustering process into a stochastic process over the configuration space in the GCA array. The proposed approach is characterized by the self-organizing clustering and many advantages in terms of the insensitivity to noise, quality robustness to clustered data, suitability for high-dimensional and massive data sets, the learning ability, and the easier hardware implementation with the VLSI systolic technology. The simulations and comparisons have shown the effectiveness and good performance of the proposed GCA approach to data clustering

Quantum Particles Model for Data Clustering in Enterprise Computing

This paper presents a new generalized quantum particle model for data self-organizing clustering. The stochastic motion and collision of quantum particles give rise to a stochastic process of quantum entanglement of particles. The stationary probability distribution over the configuration space of entangled particles results in the optimally clustering solution of the given data set. The quantum particle model has advantages in terms of the insensitivity to noise, the quality robustness to clustered data, the learning ability, and the suitability for high-dimensional multi-shape large-scale data sets. In comparison with the classical version of particle model and the cellular automata, the quantum particle mode has much faster speed and higher quality for clustering. The simulation and comparison show the effectiveness and good performance of the proposed quantum particle approach to data clustering.

Parallel Optimization Based on Generalized Cellular Automata

The fast packet switching (FPS) in computer networks. This paper further extends GCA to effectively solving a class of optimization problems subject to a binary constraint matrix, including the FPS problem and the traveling salesman problem (TSP) that is NP-hard. In contrast to Hopfleld-type neural networks and cellular neural networks, the proposed GCA approach has the pyramid architecture and evolutionary dynamics related to multi-granularity macro-cells. This paper discusses the details regarding the dynamics and properties of the improved GCA.erms of the solution quality.

Optimal Control of Network Services Based on Generalized Particle Model

The bandwidth allocation problem in ATM networks is NP-complete. This paper presents a novel generalized particle approach (GPA) to optimize the bandwidth allocation and QoS parameter for ATM networks. The GPA transforms the optimization of ATM networks into a kinematics and dynamics of numerous particles in a force-field. The GPA has many advantages in terms of the higher parallelism, multi-objective optimization, multi-type coordination, and easiness for hardware implementation. During the ATM networks optimization, the GPA may deal with a variety of random and emergent phenomena, such as the congestion, failure, and interaction. This paper also gives the GPAs properties regarding its correctness, convergency and stability. The simulations have shown the effectiveness and suitability of the GPA to the optimization of ATM networks

A Novel Quantum Particle Approach to Self-Organizing Clustering

Most of currently used approaches to data clustering are not qualified to quickly cluster a high-dimensional large-scale database. This paper is devoted to a novel generalized quantum particle model (GQPM) to data self-organizing clustering. The GQPM approach transforms the data clustering process into a stochastic process of particle motion, collision and quantum entanglement on a particle array. In comparison with the GPM clustering method we have proposed before, the GQPM has much faster speed and higher quality for clustering. GQPM is also characterized by the self-organizing clustering and has advantages in terms of the insensitivity to noise, the quality robustness to clustered data, the learning ability, the suitability for high-dimensional multi-shape large-scale data sets. The simulations and comparisons have shown the effectiveness and good performance of the proposed GQPM approach to data clustering

A generalized particle model for social coordination and autonomy in MAS

In multi-agent systems (MAS), every autonomous agent tries to increase its own personal utility under complex environment involving a variety of types of social coordination. Modeling the social coordination and autonomy of agents is of great significance for distributed problem-solving in MAS. This paper is devoted to a novel generalized particle approach to model social coordination and autonomy in MAS. At first, we analyze and formalize some typical types of social coordination in MAS. Then we discuss the generalized particle model (GPM) for distributed problem-solving in MAS that is related to social coordination and social dynamics. Finally, we demonstrate the GPM-based parallel algorithm and its properties for parallel distributed task allocation and resource assignment in MAS.