Fuzzy C-means soft balance clustering algorithm driven by distribution entropy

The invention relates to a clustering problem in the field of machine learning, in particular to a fuzzy C-means soft balance clustering algorithm driven by distribution entropy, which comprises the following steps: 1, defining the distribution entropy of a hard label matrix; 2, defining a corresponding relationship between the hard label matrix and the fuzzy membership matrix; 3, constructing a square loss term by adopting a Frobenius norm to measure the distance between the hard label matrix and the fuzzy membership matrix; 4, constructing a fuzzy C-means soft balance clustering model drivenby the distribution entropy in combination with the distribution entropy and the square loss term of the label matrix; and 5, solving the model by adopting an alternate optimization strategy. 本发明涉及机器学习领域中的聚类问题，具体涉及一种分布熵驱动的模糊C均值软平衡聚类算法，包括以下步骤：第一，定义硬标签矩阵的分布熵；第二，定义硬标签矩阵和模糊隶属度矩阵之间的对应关系；第三，采用Frobenius范数构建平方损失项度量硬标签矩阵和模糊隶属度矩阵之间的距离；第四，结合标签矩阵的分布熵和平方损失项，构建分布熵驱动的模糊C均值软平衡聚类模型；第五，采用交替优化的策略对模型进行求解。