Weighted Data Gravitation Classification for Standard and Imbalanced Data

Abstract

Gravitation is a fundamental interaction whose concept and effects applied to data classification become a novel data classification technique. The simple principle of data gravitation classification (DGC) is to classify data samples by comparing the gravitation between different classes. However, the calculation of gravitation is not a trivial problem due to the different relevance of data attributes for distance computation, the presence of noisy or irrelevant attributes, and the class imbalance problem. This paper presents a gravitation-based classification algorithm which improves previous gravitation models and overcomes some of their issues. The proposed algorithm, called DGC+, employs a matrix of weights to describe the importance of each attribute in the classification of each class, which is used to weight the distance between data samples. It improves the classification performance by considering both global and local data information, especially in decision boundaries. The proposal is evaluated and compared to other well-known instance-based classification techniques, on 35 standard and 44 imbalanced data sets. The results obtained from these experiments show the great performance of the proposed gravitation model, and they are validated using several nonparametric statistical tests.

A locally weighted learning method based on a data gravitation model for multi-target regression

“…As for classification tasks, several data gravitation-based algorithms have been proposed 32,7,46,33 . Peng et al 32 presented one of the most complete work concerning data gravitationbased classification.…”

“…Data gravitation-based models have demonstrated to be less sensitive in those cases where kNN methods severely deteriorate their performance; kNN methods tend to deteriorate their predictive performance on data with high dimensionality, non-separable classes, or non-uniform distribution of examples per classes 27 . In this sense, Cano et al 7 proposed a data gravitation model that outperforms several state-of-the-art kNN methods, and they also demonstrated its efficacy in imbalanced data. On the other hand, Reyes et al 34 presented an effective data gravitation-based algorithm to solve the multi-label learning problem, a paradigm very close to multitarget regression.…”

“…However, constructing artificial data particles from several examples have shown several disadvantages, leading to a significant degradation in the effectiveness of data gravitation-based methods 7,34 . In this work, we followed the idea pro- ¶ Data particle is a kind of data unit constructed from several training examples.…”

Gravitational Search Algorithms in Data Mining: A Survey

“…Amplified gravitation coefficient (AGC) that contains class imbalance information has been used for this purpose [38]. Weighted data gravitation classification [39] has also been used to handle the imbalance data problem. In this work, a matrix of weights that describe the importance of each attribute in the classification of each class is used to avoid the imbalance data problem.…”

Gravitation-Based Edge Detection for Hyperspectral Images

“…Accordingly, existing approaches for color or feature-based color image segmentation (SGISA) [31] and the gravitational collapse (CTCGC) approach for color texture classification [36]. Other gravitation model based approaches for classification include data gravitation classification (DGC) [37][38][39] and gravitational self-organizing maps (GSOM) [40]. The advantages of the gravity field model have inspired the development of new edge-detection method for gray-scale images [41,42].…”