Image GPT For Unsupervised Learning Without Domain Knowledge And Image Generation Is Awesome! (Representation Learning Of Images Summer 2020 Feature 1)

Image Recognition 09/07/2020

3 main points
✔️ Unknown Domain Knowledge Unsupervised Representation Learning Successful, Valuable Proof-of-Concept (PoC), Ignoring the Computational Complexity
✔️ Amazing image generation capability even as a generation model
✔️ SOTA performance with image classification using acquired representations

Generative Pretraining from Pixels
written by Mark Chen, Alec Radford, Ilya Sutskever (OpenAI)
(Submitted on 17 Jun 2020)
Comments: Accepted at ICML2020
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Paper Official Code COMM Code

In this three-part series, we will introduce various methods of unsupervised learning in a special feature entitled "Learning to Represent Images in Summer 2020". Perhaps because this field has reached a certain level of performance, many different methods have been proposed, and the game is heating up. I would like to present a selection of these methods.

The writer's special project entitled "Learning to Express Images Summer 2020" introduces various methods of unsupervised learning.

Part 1. Image GPT for domain knowledge-free and unsupervised learning, and image generation is amazing!
Part 2. Contrastive Learning's Two Leading Methods SimCLR And MoCo, And The Evolution Of Each
Part 3. SOTA With Contrastive Learning And Clustering!
Part 4. Questions For Contrastive Learning : "What Makes?"
Part 5. The Versatile And Practical DeepMind Unsupervised Learning Method

Introduction - The Significance of Unsupervised + Expressive Learning

Since this is an introduction to papers that deal primarily with images throughout this series, we'll use a dataset like CIFAR-10 to illustrate this. When we do "unsupervised learning" on a dataset like this, we can think of a case where we would be clustering without using labels at all.

To read more,

Please register with AI-SCHOLAR.