Perforated is the data-efficiency layer for machine learning that helps teams build smaller, more accurate models using less training data. By adding biological-inspired dendritic modules to artificial neurons during training, Perforated allows PyTorch models to extract more value from existing datasets.