Spatial Metric Space for Pattern Recognition Problems
Abstract
The definition of weighted distance measure involves weights. The paper proposes a weighted distance measure without the help of weights. Here, weights are intrinsically added to the measure, and for this, the concept of metric space is generalized based on a novel divided difference operator. The proposed operator is used over a two-dimensional sequence of bounded variation, and it generalizes metric space with the introduction of a multivalued metric space called spatial metric space. The environment considered for the study is a two-dimensional Atanassov intuitionistic fuzzy set (AIFS) under the assumption that membership and non-membership components are its independent variables. The weighted distance measure is proposed as a spatial distance measure in the spatial metric space. The spatial distance measure consists of three branches. In the first branch, there is a domination of membership values, non-membership values dominate the second branch, and the third branch is equidominant. The domination of membership and non-membership values are not in the form of weights in the proposed spatial distance measure, and hence it is a measure independent of weights. The proposed spatial metric space is mathematically studied, and as an implication, the spatial similarity measure is multivalued in nature. The spatial similarity measure can recognize a maximum of three patterns simultaneously. The spatial similarity measure is tested for the pattern recognition problems and the obtained classification results are compared with some other existing similarity measures to show its potency. This study connects the double sequence to the application domain via a divided difference operator for the first time while proposing a novel divided difference operator-based spatial metric space.
- Publication:
-
arXiv e-prints
- Pub Date:
- December 2023
- DOI:
- 10.48550/arXiv.2312.12020
- arXiv:
- arXiv:2312.12020
- Bibcode:
- 2023arXiv231212020D
- Keywords:
-
- Mathematics - Metric Geometry
- E-Print:
- 27