Unproportional mosaicing
This addresses data shift issues for computer vision applications, but appears incremental as it builds on existing augmentation methods.
The paper tackles the problem of data shift in machine learning by introducing Unproportional mosaicing (Unprop), a new data augmentation technique that randomly splits images into various-sized blocks and swaps their content, achieving a lower error rate when combined with other state-of-the-art augmentations.
Data shift is a gap between data distribution used for training and data distribution encountered in the real-world. Data augmentations help narrow the gap by generating new data samples, increasing data variability, and data space coverage. We present a new data augmentation: Unproportional mosaicing (Unprop). Our augmentation randomly splits an image into various-sized blocks and swaps its content (pixels) while maintaining block sizes. Our method achieves a lower error rate when combined with other state-of-the-art augmentations.