H. Yuan, Y. Ye, and J. Deng
IB method, sIB algorithm, simulated annealing, mutual information
An iterative sIB algorithm based on simulated annealing is proposed in this paper, which improves the efficiency and accuracy of the sIB algorithm. First, this algorithm chooses some positions randomly from the initial solution vector of a basic sIB algorithm, then randomly changes their corresponding clustering labels and optimizes them during an annealing processing. Experimental results on the benchmark data sets demonstrate that the proposed SA_sIB algorithm outperforms the sIB algorithm on both accuracy and efficiency.
Important Links:
Go Back
