VS-SMOTE (VSSMOTE, 2025)

Resampling with VSSMOTE.

Usage

VSSMOTE(x, y, k = 5, k1 = 0, k2 = k, ovRate = 1, ...)

Arguments

x

feature matrix or data.frame.

y

a factor class variable. Can have more than two classes.

k

number of nearest neighbors of the class that each sample is linked to (connecting lines) and size of the neighborhood of the middle points. Default is 5.

k1, k2

a connecting line is used if the number of samples of the class among the k nearest neighbors of its middle point is between k1 and k2 (high-value space). k1 = 0 and k2 = k (default) give the ordinary SMOTE. Raising k1 removes lines in majority dominated (noisy) spaces, lowering k2 removes lines in the safest spaces. If NULL, k2 is k.

ovRate

Oversampling rate. Class \(j\) receives round(ovRate * (n_max - n_j)) synthetic samples, where n_max is the size of the largest class (halves are rounded up). Default is 1 meaning full balance. A vector with one value per class is also accepted.

...

not used.

Value

a list with resampled dataset.

x_new

Resampled feature matrix.

y_new

Resampled target variable.

x_syn

Generated synthetic feature data.

y_syn

Generated synthetic label data.

Details

Year: 2025. Each sample is linked to its k nearest neighbors of the class. The middle point of every connecting line is used to judge the space of the line, and synthetic samples are generated uniformly along randomly chosen lines of the high-value spaces only. If no line qualifies, all lines are used with a warning.

Can work with classes more than 2: every class is oversampled against the union of the other classes (one-versus-rest).

References

Qiu, F., Yang, F., Xiong, Y., Chen, T., Xiao, P., Zheng, W., ... & Zhou, K. (2025). VS-SMOTE: Leveraging high-value spaces to balance noise control and data diversity for class imbalanced learning. Neurocomputing, 131321.

Author

Fatih Saglam, saglamf89@gmail.com

Examples

set.seed(1)
x <- rbind(matrix(rnorm(2000, 3, 1), ncol = 2, nrow = 1000),
           matrix(rnorm(100, 5, 1), ncol = 2, nrow = 50))
y <- as.factor(c(rep("negative", 1000), rep("positive", 50)))

m <- VSSMOTE(x = x, y = y, k = 5, k1 = 2, k2 = 5)
table(m$y_new)
#> 
#> negative positive 
#>     1000     1000 
plot(m$x_new, col = m$y_new)