Table 3 Analysis of synthetic data generation mechanisms
From: Tabular and latent space synthetic data generation: a literature review
Type | Mechanism | Smoothness | Manifold | Priv. | Reg. | Ovs. | AL | Semi-SL | Self-SL |
|---|---|---|---|---|---|---|---|---|---|
Perturbation | Random | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\times\) |
Laplace | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\times\) | \(\times\) | \(\times\) | |
Gaussian | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\checkmark\) | \(\checkmark\) | |
Swap-noise | \(\times\) | \(\times\) | \(\times\) | \(\times\) | \(\times\) | \(\times\) | \(\checkmark\) | \(\checkmark\) | |
Zero-out noise | \(\times\) | \(\times\) | \(\times\) | \(\times\) | \(\times\) | \(\times\) | \(\times\) | \(\checkmark\) | |
Gaussian Gen. | \(\times\) | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\times\) | |
Gaussian Mix. | \(\times\) | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\times\) | |
KDE | \(\times\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\times\) | |
PGM | Bayesian Net. | \(\times\) | \(\times\) | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\times\) | \(\times\) |
Gibbs | \(\times\) | \(\times\) | \(\times\) | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\times\) | |
Random Walk | \(\times\) | \(\times\) | \(\times\) | \(\times\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\times\) | |
Linear | Between-class Int. | \(\times\) | \(\checkmark\) | \(\times\) | \(\checkmark\) | \(\times\) | \(\checkmark\) | \(\checkmark\) | \(\times\) |
Within-class Int. | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\times\) | |
Extrapolation | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\times\) | |
Hard Extra. | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\times\) | |
Inter.+Extra. | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\times\) | |
Difference Transf. | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\times\) | \(\times\) | |
Geometric | Hypersphere | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\times\) |
Triangular | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\times\) | \(\times\) | \(\checkmark\) | \(\times\) | |
Hyperrectangle | \(\times\) | \(\checkmark\) | \(\times\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\times\) | \(\times\) | |
Neural nets. | GAN | \(\times\) | \(\times\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\times\) | \(\times\) |
AE | \(\times\) | \(\times\) | \(\times\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\times\) | |
Others | Exponential M. | \(\times\) | \(\times\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\times\) | \(\times\) | \(\times\) |
Reconstruction err. | \(\times\) | \(\times\) | \(\times\) | \(\times\) | \(\checkmark\) | \(\times\) | \(\times\) | \(\times\) |