Artificial Intelligence: Week #44 | 2020
This week in AI & Machine Learning: Robotaxis, AI reduces hospital admissions, Pytorch 1.7, mixed precision training, AWS re:invent, people are awesome, and more!
Artificial Intelligence News:
Voyage Partners with First Transit to Bring Robotaxis to More People
In some really exciting self driving car news, Voyage announces a partnership with First Transit. This is a really neat and different collaborative approach to deploying autonomous vehicles to more people. I won’t be too surprised if more companies end up taking a similar approach. Especially if you’re not already one of the massive tech giants like Google, Amazon, or Tesla.
Microsoft Releases Preview of Lobe Training App for Machine Learning
Microsoft released a preview of the Lobe app which promises to make machine learning easier for local model development for people without much data science experience. Check out the app at lobe.ai.
If you’re looking for more platform options making machine learning and computer vision easier, no matter your experience level, request a demo of our Sense platform!
AI Reduces Hospital Admissions
Clare Medical used artificial intelligence in a study that was able to reduce hospitalization by over 51% for at-risk patients. The AI was able to make predictions of series medical events allowing for clinicians to proactively interview. A tool like this could be the future of reducing hospital overcrowding and saving lives of at-risk populations!
Developer Tools & Education:
Pytorch 1.7 has Been Released
If you’ve been reading these weekly AI reports it will be no surprise to you that we’re a fan of Pytorch! Version 1.7 is officially released with Cuda 11 support. Check out the release notes.
Hugging Face Tokenizers 0.9 are here
Hugging Face releases Tokenizers 0.9 which includes full support for Unigram, Numpy arrays as input, new components for better customization, and shiny new docs.
Targeted Adversarial Attacks with Keras and TensorFlow
learn how to perform targeted adversarial attacks to trick convolutional neural networks and construct targeted adversarial images using Keras, TensorFlow, and Deep Learning.
Mixed Precision Training
Learn how to quickly do mixed precision training with Pytorch Lighting.
Train Conversational AI in 3 lines of code with NeMo and Lightning
Learn how to easily train a Natural Language Processing (NLP) model with the NeMo NVIDIA framework.
Online Notebooks with GPU
If you’re a fan of Jupyter Notebooks or Google Colab, check out this new player in the field of online notebooks for deep learning with GPU acceleration.
Upcoming Online AI & Data Events:
Intro to Machine Learning | Nov 10 – 5:30pm PDT
Join me for a fun introduction to machine learning and build classification and regression models with Python.
AWS re:invent | Nov. 30 – Ded. 18, 2020
AWS re:invent is going virtual and is free this year!
Autonomous Drones Saving Lives with Keenan Wyrobek, Founder, Zipline | Nov 4 – 5:00pm PDT
This event is talking about using autonomous drones to save lives. Do I really need to say anymore??
Can Consciousness be Explained Fully by science?: A debate | Nov 7 – 1:00pm PDT
A fun sounding debate around explaining consciousness with science–a common topic in the field of artificial general intelligence.
Interesting Podcasts & Interviews:
It seems like all the good AI & machine learning podcasts came out last week. If you found one you enjoyed, share it with me and I can add it here!
The Future of Education and AI with Sal Khan
Sal is the founder of the popular Khan Academy. In this discussion, hear about his thoughts on machine learning and AI being used broadly in education and much more.
“People are Awesome: Why You Shouldn’t Fear Public Speaking” with Sage Elliott of Sixgill
Last week I was interviewed for an Educative Sessions podcast. It wasn’t entirely related to AI, but I talk about how I got started doing technical talks and public speaking. If you ever have any questions about getting started in computer vision or giving your first talk reach out to me!
Notable Research Papers:
Some of the interesting machine learning papers published this week.
COG: Connecting New Skills to Past Experience with Offline Reinforcement Learning
Online Semi-Supervised Learning with Bandit Feedback
Object-aware Feature Aggregation for Video Object Detection
GiBERT: Introducing Linguistic Knowledge into BERT through a Lightweight Gated Injection Method
About Sixgill:
Sixgill, LLC provides custom enterprise AI solutions, end-to-end machine learning lifecycle management, and fast data annotation for computer vision.
