The biclustering algorithms are a kind of new data mining methods, which are commonly evaluated with mean squared residue. Biclustering algorithms based on mean squared residue mostly use a greedy strategy, which can not obtain accurate clusters with appropriate size. However, fuzzy theory can improve the performance of coclustering algorithms based on mean squared residue clustering and obtain more accurate clusters with appropriate size. This paper presents a fuzzy biclustering algorithm for solving a single cluster based on fuzzy theory to improve the performance of biclustering algorithms based on mean squared residue clustering. Firstly, the paper defines the fuzzy variables named significant indicators for biclustering problem. Then, this paper builds a novel fuzzy biclustering model, and gives an algorithm and its convergence analysis. Finally, compared with the biclustering algorithm FLOC and the fuzzy coclustering simulation data and real data, the fuzzy biclustering algorithm is more effective.%双聚类算法是一类新型数据挖掘聚类算法,通常以均方残差为评价指标.基于均方残差的双聚类算法,大多采用贪婪策略求解,通常不能得到大小适中且结果准确的簇.而在联合聚类中,模糊理论能改善这种基于均方残差的算法,得到大小适中且结果准确的簇.为了提高基于均方残差双聚类算法的性能,本文结合模糊理论提出一种求解单一簇的模糊双聚类算法.首先,提出定义双聚类簇内的模糊变量,即显著性指标；然后,建立基于显著性指标的模糊双聚类模型,并给出算法及其收敛性分析；最后,利用仿真数据和真实数据,将模糊双聚类算法与FLOC双聚类算法和模糊联合聚类算法进行对比,以验证模糊双聚类算法的有效性.
展开▼
