论文部分内容阅读
均场退火方法既可以看作是一种新的神经网络计算模型 ,又可视为是对模拟退火的重大改进 .该文把具有相邻约束的多层通孔最小化问题转换为更具广泛意义的 k-着色问题 ,并提出了 k-着色问题的均场退火求解算法 .算法在线段相交图模型的基础上 ,提出了相邻矩阵和交叠矩阵等概念 ,并利用换位矩阵 ,将问题映射为相应的神经网络 ,再构造了该问题的能量函数 .能量函数中的目标项、违背交叠约束的惩罚项、违背相邻约束的惩罚项和神经元归一化处理保证了网络能够求解到一个合法解 .实验结果表明 ,这是一个有效的算法 .
The mean-field annealing method can be regarded as a new neural network calculation model, which can be considered as a significant improvement on simulated annealing.This paper transforms the minimization problem of multilayer vias with adjacent constraints into a more extensive And proposes a mean-field annealing algorithm for solving the k-coloring problem.On the basis of the line segment intersection graph model, the concepts of adjacent matrix and overlapping matrix are proposed and the transposition matrix The problem is mapped to the corresponding neural network and the energy function of the problem is reconstructed.The target items in the energy function, the penalty items that violate the overlap constraint, the penalty items that violate the adjacent constraints and the neuron normalization process ensure that the network can Solve a legal solution. Experimental results show that this is an effective algorithm.