- 期刊
Exploring Effectiveness of Interval Type-2 Fuzzy K-Nearest Neighbor Classifier in Different Distance Metric Spaces
摘要
Type-2 fuzzy sets characterized by three-dimensional membership functions that are themselves fuzzy have been attracting many interests in recent years. Type-2 fuzzy sets can simultaneously handle randomness and fuzziness until final decision-making. Fuzzy K-nearest neighbor classifier is commonly used for pattern classification, and it generally performs on the Euclidean metric space. In this study, we will explore the effectiveness of an addressed interval type-2 fuzzy K-nearest neighbor classifier (IT2FKNNC) in four distance metric spaces. Three benchmark hyperspectral image datasets with hundreds of dimensions are employed to evaluate the performance of some KNN type classifiers. The experimental results show that the IT2FKNNC outperforms other classifiers in the four distance spaces. Importantly, we find that classifiers perform better in the cosine and cityblock distance spaces than in the Euclidean distance space for high-dimensional data classification.
延伸閱讀
- Chu, C. Y. (2018). Robust Fuzzy C-Means Clustering Algorithm for Interval Data [master's thesis, Chung Yuan Christian University]. Airiti Library. https://doi.org/10.6840/cycu201800251
- 彭得裕（2021）。Study on Signal Identification by Applying the K-Nearest Neighbors Algorithm〔碩士論文，健行科技大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0022-0604202113383700
- 李來錫、蔡耀德（2006）。Searching the Nearest Neighbor in Collaborative Filtering Systems using the Fuzzy Linguistic Analysis。管理科學研究，3(2)，75-86。https://doi.org/10.6873/MSR.200612_3(2).0005
- Barigou, F. (2016). IMPROVING K-NEAREST NEIGHBOR EFFICIENCY FOR TEXT CATEGORIZATION. Neural Network World, 26(1), 45-65. https://doi.org/10.14311/NNW.2016.26.003
- Li, C. L. (2010). Top-N Query Processing on Spatial Databases Considering Bi-chromatic Reverse k-Nearest Neighbors [master's thesis, National Tsing Hua University]. Airiti Library. https://doi.org/10.6843/NTHU.2010.00466