In this paper, we present an artificial life method of the cellular neural network for max-clique problem. The method is intended to provide an optimum parallel algorithm for solving the max-clique problem. To do this...
详细信息
In this paper, we present an artificial life method of the cellular neural network for max-clique problem. The method is intended to provide an optimum parallel algorithm for solving the max-clique problem. To do this we use the cellular neural network to get a maximum clique. Some of the instances are simulated to verify the proposed method with the simulation results showing that the solution quality is superior to that of best existing parallel algorithm. We also test the learning method on total coloring problem.
Artificial neural network (ANN) is widely used in pattern recognition related area. In some case, the computational load is very heavy, in other case, real time process is required. So there is a need to apply a paral...
详细信息
Artificial neural network (ANN) is widely used in pattern recognition related area. In some case, the computational load is very heavy, in other case, real time process is required. So there is a need to apply a parallel algorithm on it, and usually the computation for ANN is inherently parallel. In this paper, graphic hardware is used to speed up the computation of ANN. In recent years, graphic processing unit (GPU) grows faster than CPU. Graphic hardware venders provide programmability on GPU. In this paper, application of commodity available GPU for two kinds of ANN models was explored. One is the self-organizing maps (SOM); the other is multi layer perceptron (MLP). The computation result shows that ANN computing on GPU is much faster than on standard CPU when the neural network is large. And some design rules for improve the efficiency on GPU are given.
The Key Message (KM) approach prioritizes communications along the critical path to speed up the execution of parallel applications in a cluster environment. In this article, we introduce a family of three dynamic KM ...
详细信息
The Key Message (KM) approach prioritizes communications along the critical path to speed up the execution of parallel applications in a cluster environment. In this article, we introduce a family of three dynamic KM algorithms that take into account of the dynamic nature of underlying networks to improve over their predecessor. These three dynamic KM algorithms are: Fixed Critical Path Dynamic KM (FDKM), Centralized Predictor Dynamic KM (CDKM), and Distributed Dynamic KM (DDKM). Our performance study based on simulation showed that in general CDKM performs better and is more stable than FDKM and DDKM under heavy traffic conditions. The performance of DDKM is close to that of CDKM and better than that of FDKM on average. By taking advantages of both FDKM and CDKM, DDKM can be expected to be a promising optimization in a real cluster environment.
In this paper we introduce a new algorithm for computing near optimal schedules for task graph scheduling problems. In contrast to conventional approaches for solving those scheduling problems, our algorithm is based ...
详细信息
In this paper we introduce a new algorithm for computing near optimal schedules for task graph scheduling problems. In contrast to conventional approaches for solving those scheduling problems, our algorithm is based on the same principles that ants use to find shortest paths between their nest and food sources. Like their natural counterparts, artificial ants cooperate by means of pheromone trails where information about the quality of the possible solution's building blocks is stored. Based on this common communication structure, new solutions emerge by means of cooperative interaction between the ants. In the paper we demonstrate how this basic principle can be adapted to solve scheduling problems. We also evaluated the performance of the proposed ANTLS-algorithm (ant list scheduler) by means of a comprehensive test bench with more than 30,000 test cases. Compared to two conventional and two other nature-inspired approaches it performed very well.
In this paper we propose a constant-time parallel algorithm for implementing the message-passing decoder of LDPC codes on a two dimensional R-Mesh, trying to keep the number of processors small. The R-Mesh provides dy...
详细信息
In this paper we propose a constant-time parallel algorithm for implementing the message-passing decoder of LDPC codes on a two dimensional R-Mesh, trying to keep the number of processors small. The R-Mesh provides dynamic reconfiguration, hardware reuse, and flexibility to problem changes. To decode a different code, we may simply set up the required connections between the bit-nodes and check-nodes by modifying the initialization phase of the R-Mesh algorithm. No extra wiring or hardware changes are required, as compared to other existing approaches. Moreover, the same hardware can implement the decoder in both probability and logarithm domains. We illustrate that the R-Mesh is an efficient model for parallel implementation of the decoder in terms of time complexity, flexibility to problem changes and simplicity of routing messages.
This paper proposes a view-oriented update protocol with integrated diff for efficient implementation of a view-based consistency model which supports a novel view-oriented parallel programming style based on distribu...
详细信息
ISBN:
(纸本)9780780390744
This paper proposes a view-oriented update protocol with integrated diff for efficient implementation of a view-based consistency model which supports a novel view-oriented parallel programming style based on distributed shared memory. View-oriented parallel programming requires the programmer to divide the shared data into views according to the nature of the parallel algorithm and its memory access pattern. The advantage of this programming style is that it offers the potential for the underlying distributed shared memory system to optimize consistency maintenance. The View-oriented update protocol with integrated diff is proposed to exploit this performance potential. This protocol is compared with a traditional diff-based protocol and an existing home-based protocol. Experimental results demonstrate that the performance of the proposed protocol is significantly better than the diff-based protocol and the home-based protocol.
In this paper we study the complexity of bounded color multiplicity graph isomorphism BCGI/sub b/: the input is a pair of vertex-colored graphs such that the number of vertices of a given color in an input graph is bo...
详细信息
In this paper we study the complexity of bounded color multiplicity graph isomorphism BCGI/sub b/: the input is a pair of vertex-colored graphs such that the number of vertices of a given color in an input graph is bounded by b. We show that BCGI/sub b/ is in the #L hierarchy (more precisely, the Mod/sub k/L hierarchy for some constant k depending on b). Combined with the fact that bounded color multiplicity graph isomorphism is logspace many-one hard for every set in the Mod/sub k/L hierarchy for any constant k, we get a tight classification of the problem using logspace-bounded counting classes.
We present distributed algorithms for sensor localization based on the Gauss-Newton method. Each sensor updates its estimated location by computing the Gauss-Newton step for a local cost function and choosing a proper...
详细信息
We present distributed algorithms for sensor localization based on the Gauss-Newton method. Each sensor updates its estimated location by computing the Gauss-Newton step for a local cost function and choosing a proper step length. Then it transmits the updated estimate to all the neighboring sensors. The proposed algorithms provide non-increasing values of a global cost function. It is shown in the paper that the algorithms have computational complexity of O(n) per iteration and a reduced communication cost over centralized algorithms.
Many parallel algorithms have been proposed for computing the two-dimensional block-based medial axis transform (2D/spl ***/BB/spl ***/MAT). Unfortunately, almost all of them are dedicated for solving 2D/spl ***/BB/sp...
详细信息
Many parallel algorithms have been proposed for computing the two-dimensional block-based medial axis transform (2D/spl ***/BB/spl ***/MAT). Unfortunately, almost all of them are dedicated for solving 2D/spl ***/BB/spl ***/MAT problem and are very difficult to be extended for solving the 3D/spl ***/BB/spl ***/MAT problem. In this paper, an O(1) time algorithm for solving the 2D/spl ***/BB/spl ***/MAT of a binary image P of size N /spl times/ N on an LARPBS is first developed. The running time of this algorithm has a smaller constant factor compared with those of all the other previous proposed O(1) time algorithms. Then this algorithm is extended for computing the 3D/spl ***/BB/spl ***/MAT of a binary image V of size N /spl times/ N /spl times/ N in O(1) time on an LARPBS. The running time of this algorithm has a small constant factor also. To the best of our knowledge, this is the first parallel algorithm proposed for solving the 3D/spl ***/BB/spl ***/MAT problem known.
暂无评论