Your Daily AI Research tl;dr — 2022–08–22 🧠

Enhancing Diffusion-Based Image Synthesis, Cold Diffusion and Google’s language model helping robots!

Welcome to your official daily AI research tl;dr (often with code and news) for AI professionals where I share the most exciting papers I find daily, along with a one-liner summary to help you quickly determine if the article (and code) is worth investigating.

1️⃣ Enhancing Diffusion-Based Image Synthesis with Robust Classifier Guidance

They define and train a time-dependent adversarially robust classifier and use it as guidance for a generative diffusion model based on the recent discoveries of adversarially robust classifiers exhibiting gradients that are aligned with human perception.

Link to the paper: https://arxiv.org/pdf/2208.08664.pdf

2️⃣ Cold Diffusion: Inverting Arbitrary Image Transforms Without Noise

“While standard diffusions are built on Gaussian noise, we show that generative models can be built on arbitrary and even noiseless/cold image transforms, including the ImageNet-C snowification operator, and an animorphosis operator that adds a random animal image from…