XCiT: Cross-Covariance Image Transformers (Facebook AI Machine Learning Research Paper Explained)

#xcit #transformer #attentionmechanism

After dominating Natural Language Processing, Transformers have taken over Computer Vision recently with the advent of Vision Transformers. However, the attention mechanism's quadratic complexity in the number of tokens means that Transformers do not scale well to high-resolution images. XCiT is a new Transformer architecture, containing XCA, a transposed version of attention, reducing the complexity from quadratic to linear, and at least on image data, it appears to perform on par with other models. What does this mean for the field? Is this even a transformer? What really matters in deep learning?

OUTLINE: 0:00 - Intro & Overview 3:45 - Self-Attention vs Cross-Covariance Attention (XCA) 19:55 - Cross-Covariance Image Transformer (XCiT) Architecture 26:00 - Theoretical & Engineering considerations 30:40 - Experimental Results 33:20 - Comments & Conclusion

Paper: https://arxiv.org/abs/2106.09681 Code: https://github.com/facebookresearch/xcit

Abstract: Following their success in natural language processing, transformers have recently shown much promise for computer vision. The self-attention operation underlying transformers yields global interactions between all tokens ,i.e. words or image patches, and enables flexible modelling of image data beyond the local interactions of convolutions. This flexibility, however, comes with a quadratic complexity in time and memory, hindering application to long sequences and high-resolution images. We propose a "transposed" version of self-attention that operates across feature channels rather than tokens, where the interactions are based on the cross-covariance matrix between keys and queries. The resulting cross-covariance attention (XCA) has linear complexity in the number of tokens, and allows efficient processing of high-resolution images. Our cross-covariance image transformer (XCiT) is built upon XCA. It combines the accuracy of conventional transformers with the scalability of convolutional architectures. We validate the effectiveness and generality of XCiT by reporting excellent results on multiple vision benchmarks, including image classification and self-supervised feature learning on ImageNet-1k, object detection and instance segmentation on COCO, and semantic segmentation on ADE20k.

Authors: Alaaeldin El-Nouby, Hugo Touvron, Mathilde Caron, Piotr Bojanowski, Matthijs Douze, Armand Joulin, Ivan Laptev, Natalia Neverova, Gabriel Synnaeve, Jakob Verbeek, Hervé Jegou

Links: TabNine Code Completion (Referral): http://bit.ly/tabnine-yannick YouTube: https://www.youtube.com/c/yannickilcher Twitter: https://twitter.com/ykilcher Discord: https://discord.gg/4H8xxDF BitChute: https://www.bitchute.com/channel/yannic-kilcher Minds: https://www.minds.com/ykilcher Parler ... https://www.youtube.com/watch?v=g08NkNWmZTA