AugMax Explained!

There has been a new breakthrough in using Data Augmentation in Deep Learning. Researchers from NVIDIA, Caltech, UT Austin, and Arizona State have published AugMax. The acronym AugMax communicates the use of Data Augmentation and Maximizing the Diversity and Difficulty of Augmented Examples. AugMax is basically free lunch in Deep Learning meaning that it is a very strong technique to add to any existing Deep Learning workflow. But more particularly, Data Augmentation has been heavily studied with Image Data in Computer Vision. AugMax is the sequel to AugMix. This predecessor tree of research in Data Augmentation is roughly AutoAugment to RandAugment to AugMix, and now AugMax. AugMax further utilizes Friendly Adversarial Training -- rather than using an adversarial search that augments images to be as challenging as possible, the authors utilize an Entropy proxy to better ensure that these adversarial images aren’t just static noise maps. Further, AugMax utilizes a new normalization scheme. From StyleGAN to GauGAN, researchers at NVIDIA have made great use of these normalization layers and they didn’t disappoint in this one. Solving Robustness and Distribution Shift wlll be huge for the trust-worthiness of Deep Learning, as well as all the applications highlighted in the WILDS benchmark -- which I highly recommend checking out if interested in Deep Learning research. Links: Weaviate Web Demo: https://console.semi.technology/ Henry AI Labs Weaviate Query Tutorial: https://www.youtube.com/watch?v=K_2X48Tln9U You might think this is cool as well - Mathcha.io (LaTex Equation Writing Tool) - https://www.mathcha.io/editor AugMax Paper Link: https://arxiv.org/pdf/2110.13771.pdf AugMax Code Repo: https://github.com/VITA-Group/AugMax AugMix Paper Link: https://arxiv.org/pdf/1912.02781.pdf Attacks Which Do Not Kill Training Make Adversarial Learning Stronger: https://arxiv.org/pdf/2002.11242.pdf Explaining and Harnessing Adversarial Examples: https://arxiv.org/pdf/141