This week in AI & Machine Learning: AI improves video streaming, making whiskey taste better, python 3.9 released, build an object detector, creating art with robots, and more.

Artificial Intelligence News:

NVIDIA Uses Generative Adversarial Networks (GANs) to Improve Video Streaming

NVIDIA announced MAXINE, a cloud AI video streaming platform. One of the most interesting features is the ability to greatly reduce bandwidth needed for high quality video streaming. MAXINE does this by utilizing GANs to reconstruct faces with small amounts of data. With video streaming being utilized more than ever, this breakthrough can bring the streaming services to more parts of the world and save on costly data bills.

Making Whiskey Taste Better with Machine Learning

Researchers at Virginia Tech are using machine learning and natural language processing (NLP) to better understand flavor descriptions. It turns out that accurately describing tastes is a very hard problem since many flavors are subjective. Using thousands of reviews combined with NLP they hope to gain insight about how to more accurately describe flavors. I’ll be looking forward to reading the Ph.D dissertation so I can find a new favorite whiskey. 

AWS Partners with Pittsburgh Health Data Alliance to Advance AI in Healthcare  

The new partnership between Pittsburgh Health Data Alliance and Amazon Web Services (AWS) looks to advance machine learning to improve healthcare screenings for both cancer and depression.

Developer Tools & Education:

Python 3.9 is here!

On Monday we saw the official release of python 3.9. You can check out a list of the features here. I think the new dictionary merging and update operators are pretty cool!

Pytorch Lightning Team Launches

The amazing team behind pytorch lighting just raised an $18.6M series A to continue working on to make machine learning easier. 

Build an object detector with your own dataset

In this recording of a live workshop, I’ll introduce you to computer vision concepts, guide you in labeling your own dataset for almost any model, and cover training a model for object detection with detectron2.

Interesting Podcasts & Interviews:

Art(ificial) Intelligence: Pindar Van Arman Builds Robots that Paint – Ep. 129

Pinder really does put the “art” in artificial intelligence in this episode from NVIDIA’s AI podcast. Check out some of the amazing art he creates with robots and AI at

Lisa Feldman Barrett: Counterintuitive Ideas About How the Brain Works | Lex Fridman Podcast #129

Lisa Feldman Barrett is a neuroscientist, psychologist, and the author of “How Emotions Are Made”. Listen to her conversation with Lex Fridman to learn more about how the brain works and how some of these ideas can be used in the artificial intelligence field. 

How Deep Learning has Revolutionized OCR with Cha Zhang

In this episode of This Week in Machine Learning (TWiML), learn about how deep learning is revolutionizing the field of optical character recognition (OCR). 

Notable Papers:

Some of the interesting machine learning papers published this week.

NeurIPS published their list of accepted papers for 2020.

Detecting Attackable Sentences in Arguments

Understanding Clinical Trial Reports: Extracting Medical Entities and Their Relations

“LazImpa”: Lazy and Impatient neural agents learn to communicate efficiently

Self-training Improves Pre-training for Natural Language Understanding

Explaining Deep Neural Networks

BERTering RAMS: What and How Much does BERT Already Know About Event Arguments? — A Study on the RAMS Dataset

Leakage-Adjusted Simulatability: Can Models Generate Non-Trivial Explanations of Their Behavior in Natural Language?

DART: A Lightweight Quality-Suggestive Data-to-Text Annotation Tool

A Survey On Anti-Spoofing Methods For Face Recognition with RGB Cameras of Generic Consumer Devices

Automatic generation of reviews of scientific papers

 View All Blogs