Paper improves classification accuracy using synthetic data and probabilistic models.
problem Low classification accuracy in supervised learning algorithms.
method Using probabilistic mixture models to identify sub-labels and generate synthetic data.
result Better classification accuracy achieved through synthetic data generation.