SOMO comprises three major stages: Initially a Self-Organizing Map produces a two-dimensional representation of the original, usually high-dimensional, space. In this paper, we present a new oversampling method, Self-Organizing Map-based Oversampling (SOMO), which through the application of a Self-Organizing Map produces a two dimensional representation of the input space, allowing for an effective generation of artificial data points. Specifically, they generate artificial data that can be used by any algorithm, not constraining the options of the user. Although there are different strategies to tackle this problem, methods that address the problem through the generation of artificial data constitute a more general approach compared to algorithmic modifications. Learning from imbalanced datasets is challenging for standard algorithms, as they are designed to work with balanced class distributions.

An implementation 1 is made available in the Python programming language.read more read lessĪbstract: New method for generating artificial data using self-organizing maps.Provides a simple and safe way to deal with imbalanced datasets.Generates within-cluster and between cluster synthetic samples.Improves performance of classifiers and outperforms various oversampling methods. Moreover, k-means SMOTE consistently outperforms other popular oversampling methods. Empirical results of extensive experiments with 90 datasets show that training data oversampled with the proposed method improves classification results. This work presents a simple and effective oversampling method based on k-means clustering and SMOTE (synthetic minority oversampling technique), which avoids the generation of noise and effectively overcomes imbalances between and within classes. Many algorithms have been proposed for this task, but most are complex and tend to generate unnecessary noise. Such techniques, called oversamplers, modify the training data, allowing any classifier to be used with class-imbalanced datasets. While different strategies exist to tackle this problem, methods which generate artificial data to achieve a balanced class distribution are more versatile than modifications to the classification algorithm. We present empirical results that show a significant improvement in the quality of the generated data when cGAN is used as an oversampling algorithm.read more read lessĪbstract: Learning from class-imbalanced data continues to be a common and challenging problem in supervised learning as standard classification algorithms are designed to handle balanced class distributions. The performance of cGAN is compared against multiple standard oversampling algorithms. Contrary to these algorithms, in this paper the conditional version of Generative Adversarial Networks (cGAN) is used to approximate the true data distribution and generate data for the minority class of various imbalanced datasets. Therefore, these approaches are based on local information, rather on the overall minority class distribution. Standard oversampling methods are variations of the SMOTE algorithm, which generates synthetic samples along the line segment that joins minority class samples. Although there are different strategies to address this problem, methods that generate artificial data for the minority class constitute a more general approach compared to algorithmic modifications. Learning from imbalanced datasets is a frequent but challenging task for standard classification algorithms. Abstract: Application of conditional Generative Adversarial Networks as oversampling method.Generates minority class samples by recovering the training data distribution.Outperforms various standard oversampling algorithms.Performance advantage of the proposed method remains stable with higher imbalance ratios.

Smote data creator