研究目的
To find the community structure of Bayesian networks by proposing an improved K-means algorithm that incorporates mutual information into the K-means algorithm and modifies the iteration conditions.
研究成果
The improved K-means algorithm proposed in this paper can obtain the same accurate community structure in a shorter time than the FastNewman algorithm. Future work includes adding heuristic methods to determine the number of communities to improve the algorithm's performance.
研究不足
The algorithm's performance is sensitive to the number of communities and sample size, requiring careful selection for optimal results. The study also notes that for small networks with strong internal connections, the algorithm's accuracy may be lower due to the uncertainty in node ownership.
1：Experimental Design and Method Selection:
The study proposes an improved K-means algorithm (IKM algorithm) for finding community structures in Bayesian networks by incorporating mutual information and modifying iteration conditions.
2：Sample Selection and Data Sources:
Four representative networks (Sachs network, Boerlage92 network, Insurance network, and Alarm network) from the BN repository library were selected for testing.
3：List of Experimental Equipment and Materials:
Not explicitly mentioned.
4：Experimental Procedures and Operational Workflow:
The IKM algorithm involves initialization, calculation of mutual information, classification into clusters, selection of new central points, and iteration until convergence.
5：Data Analysis Methods:
Modularity Q is used to quantify the level of clustering, and the performance of the IKM algorithm is compared with the FastNewman algorithm based on Q values and running time.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
