A novel adaptive hit-distance for pattern recognition

作者：Zhen Lou^*; Zhong Jin

来源：Chinese Journal of Electronics, 2006, 15(4A): 793-796.

• biomimetic pattern recognition
• machine learning
• nearest neighbor classifier
• hit-distance
• pattern recognition

The better the matter is understood, the better the distance between any given point and the matter subspace can be defined, and vice verse. In this paper, a novel idea of distance, Hit-Distance, was proposed to generalize the representational capacity of available prototypes. It is more reasonable to utilize the proposed hit-distance to describe the distance between any given point and any matter subspace. The effectiveness of the proposed distance was indirectly evaluated by some experiments of matter classifications. Experiments were performed on 8 benchmark datasets from the UCI Machine Learning Repository. It was shown that the hit-distance based classifiers performed much better than the classical nearest neighbor classifier (NN), the Nearest feature line method (NFL) and the Nearest feature plane method (NFP).

• 出版日期2006-10
• 单位南京大学