Unsupervised Learning Based On Fuzzy Clustering

Learning is the hallmark of human intelligence, the basic means to get human intelligence, and an important brainpower act. The learning quality is the most important indicator of quality of a learning system in machine learning. Fuzzy Clustering Analysis is one of un supervised learning methods of machine learning, which has become a hot topic of machine learning research and provides an excellent means to enhance and improve the performance of machine learning technology. The research about Fuzzy clustering algorithm has great theoretical and practical significance to enhance research and improve the learning performance.As FCM algorithm has merits like simple and visual geometry in Fuzzy Clustering algorithms, so it has been applied successfully in many areas. But because based on the traditional function of the climbers FCM iterative technique to find the optimal solution, a local search algorithm essentially, so there are two fatal problems: one is slowing while dealing with a large amount of data, the second is sensitizing to the initialization of data and easily getting into local minimum.In this regard, this paper focuses on the following ways to improve on the FCM algorithm using compared and experimental methods:1. To speed up FCM clustering velocity in large amounts of data, reduce the number of iterations required for convergence and the computing time by combining several random sampling data cluster with data reduction; to improve the FCM accuracy rate, we choice characteristics by weighing their contributions to pattern classification and get better result.2. Solving the problem that FCM is prone to fall into the local optimization using genetic algorithm. A combination of genetic algorithms and FCM creates fuzzy c-means clustering algorithm GFCM based on genetic algorithms, which can get better result and raise accuracy of clustering for making full use of FCM local search virtue and genetic algorithms' ability of global search.3. As to the defect that FCM processes large amounts of data slowly, the article use neural network technology to improve fuzzy clustering algorithm. Creating the fuzzy c-means clustering algorithm FKCN based on SOM and utilizes parallel computing of SOM to enhance the speed of the clustering algorithm. The experimental results show that the algorithm is effective.The above improvements about FCM algorithm compensate for the limitations of FCM, and make the algorithm more reasonable for reducing the clustering time and raising the clustering effective. Thereby, the unsupervised learning's ability, efficiency and stability are raised and the performance of machine learning is optimized.