This week in AI & Machine Learning: OpenAI: Creating Images from Text, baking ML recipes, the future of farming, JupyterLab 3.0, Google’s Machine Learning Crash Course, Trends in Computer Vision, and more.
Artificial Intelligence News:
Named after Salvador Dalí and WALL·E, OpenAI’s new transformer language model DALL·E is offering some mind blowing image generation results from taking a few english words as parameters. The capabilities of controlling attributes, drawing multiple objects, understanding perspective, inferring context, and even visualizing internal or external structures is indeed incredibly impressive. I highly recommend reading more about this approach and playing around with inputs on the official blog post.
Read about some of the current interesting areas of research around applications of AI and farming, why the agriculture industry may be behind the curve, and how the University of Florida plans to move AI and agriculture ahead.
I think our potential new AI overlords are figuring out ways to make sure we don’t rebel by creating delicious(?) new foods? These results actually come from “breaking” a text-based receipt classifier in Google’s AutoML table tool. I am always so intrigued when AI inspires us to make or learn something new.
Developer Tools & Education:
Google launched a new machine learning crash course with TensorFlow APIs, real-world case studies, and about 15 hours of content.
I don’t know who wanted this… But, whoever you are, you can now use jupyter notebooks inside excel.
The 3.0 release of JupyterLab brings many new features to users and substantial improvements to the extension system. Check out the blog post for a list of all the features.
The latest pyimagesearch post continues the series on ArUco markers. This time, utilizing them to create augmented reality applications in OpenCV.
Hands-On Image Generation with TensorFlow: A practical guide to generating images and videos using deep learning
I’ve been excited for this book to come out! It has some really great examples of how to build image generation applications with deep learning. You can buy a copy here, and checkout the github repo here.
Interesting Podcasts & Interviews:
Listen to Dan Kokotov (VP of Engineering at Rev.ai) talk about automatic speech recognition, translation, creating products that people love, and more.
Listen to Charles Isbell & Michael Littman discuss machine learning vs. statistics, NeurIPS vs ICML, what data is more important than algorithms, and much more.
This episode of TWiML AI rewind reviews the trends happening in the past year for computer vision and provides some predictions for 2021.
This episode of TWiML AI rewind reviews the trends happening in the past year for natural language processing and provides some predictions for 2021.
This episode of TWiML AI rewind reviews the trends happening in the past year for reinforcement learning and provides some predictions for 2021.
Notable Research Papers: