Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper critiques the long-standing practice of using synthetic minority data to address class imbalance in machine learning, revealing that traditional validity checks are fundamentally flawed. By introducing a de-biased estimator that evaluates synthetic points against withheld real data, the authors demonstrate that existing methods often underestimate the invalidity of synthetic samples, with 96-99% of methods failing to provide meaningful improvements. The findings indicate that the validity of synthetic data is inherently tied to the underlying data distribution, suggesting that oversampling is unnecessary when classes are distinct and detrimental when they overlap.
Synthetic minority data may actually do more harm than good, with 96-99% of methods failing to prove their validity against real-world data.
For two decades, the standard remedy for class-imbalanced learning has been to fabricate synthetic minority examples, and the standard evidence of their validity has been a check that cannot fail: synthetic points are scored against the very data that generated them. We de-bias the check. Validity becomes a population quantity -- the probability that a synthetic point truly belongs to the minority class -- with a consistent estimator that scores synthetic points against withheld real data. Where held-out ground truth is available, the classical test underestimates true invalidity in 96-99% of method-by-imbalance-ratio cells, while the de-biased estimator tracks it closely. We prove validity is a property of the data, not the method: class overlap sets an invalidity floor no faithful generator escapes, making oversampling redundant where classes separate and invalid where they overlap. Across 91 methods, three classifiers, and datasets spanning medicine and finance -- including a generator engineered to pass the classical check -- none clears both bars: gains over the best trivial baseline are noise-thin (median below 0.01 F1, a decision threshold's reach), and most damage calibration. We release the audit as a pip-installable test and flip the burden of proof: synthetic minority data must now demonstrate, on the data at hand, both validity and information gain.