AlignMix: Improving representations by interpolating aligned features

Shashanka Venkataramanan, Ewa Kijak, laurent amsaleg, Yannis Avrithis

Abstract:

Mixup is a powerful data augmentation method that interpolates between two or more examples in the input or feature space and between the corresponding target labels. However, how to best interpolate images is not well defined. Recent mixup methods overlay or cut-and-paste two or more objects into one image, which needs care in selecting regions. Mixup has also been connected to autoencoders, because often autoencoders generate an image that continuously deforms into another. However, such images are typically of low quality.

In this work, we revisit mixup from the deformation perspective and introduce AlignMix, where we geometrically align two images in the feature space. The correspondences allow us to interpolate between two sets of features, while keeping the locations of one set. Interestingly, this retains mostly the geometry or pose of one image and the appearance or texture of the other. We also show that an autoencoder can still improve representation learning under mixup, without the classifier ever seeing decoded images. AlignMix outperforms state-of-the-art mixup methods on five different benchmarks.

One-sentence Summary: We introduce a novel mixup operation, called AlignMix where, we investigate geometric alignment for mixup based on explicit semantic correspondences in the feature space.