Unsupervised Transformation Learning via Convex Relaxations
This work addresses the challenge of learning image transformations without supervision, which could benefit applications in image editing and data augmentation, though it appears incremental as it builds on linear methods without introducing a new paradigm.
The paper tackles the problem of extracting meaningful transformations from raw images, such as altering line thickness in handwriting or lighting in portraits, by proposing an unsupervised approach that reconstructs images from linear combinations of transformed nearest neighbors, resulting in visually high-quality modified images on datasets like handwritten digits and celebrity portraits.
Our goal is to extract meaningful transformations from raw images, such as varying the thickness of lines in handwriting or the lighting in a portrait. We propose an unsupervised approach to learn such transformations by attempting to reconstruct an image from a linear combination of transformations of its nearest neighbors. On handwritten digits and celebrity portraits, we show that even with linear transformations, our method generates visually high-quality modified images. Moreover, since our method is semiparametric and does not model the data distribution, the learned transformations extrapolate off the training data and can be applied to new types of images.