适合大规模数据集且基于LLM的0阶TSK模糊分类器

针对传统分类器的泛化性能差、可解释性及学习效率低等问题, 提出0阶TSK-FC模糊分类器.为了将该分类器 应用到大规模数据的分类中, 提出增量式0阶TSK-IFC模糊分类器, 采用增量式模糊聚类算 法(IFCM($c+p$))训练模糊规则参数并通过适当的矩阵变换提升参数学习效率.仿真实验表明, 与FCPM-IRLS模糊分类器、径向基函数神经网 络相比, 所提出的模糊分类器在不同规模数据集中均能保持很好的性能, 且TSK-IFC模糊分类器在大规模数据分类中尤为突出.

Zero-order TSK fuzzy classifier based on LLM for large-scale data sets

In order to overcome the shortcoming that traditional classifiers cannot achieve satisfactory generalization performance, good interpretability and fast learning efficiency for datasets, the zero-order TSK fuzzy classifier called TSK-FC is proposed to solve the classification problem of middle-scale datasets.In order to make the TSK-FC suitable for large-scale data sets, its incremental version called TSK-IFC is developed, in which the incremental fuzzy clustering algorithm called incremental fuzzy ($c+p$)-means clustering(IFCM($c+p$)) is used to train antecedent parameters of fuzzy rules while fast consequent parameter learning is achieved through an appropriate matrix computation trick for the least learning machine.The proposed fuzzy classifiers, the TSK-FC and the TSK-IFC are experimentally compared with the conventional fuzzy classifier called FCPM-IRLS and the RBF neural network, and the results show the power of the proposed fuzzy classifiers, especially the great applicability of the TSK-IFC for large-scale data sets.