Last Week in AI
The latest developments in AI research and technology.
Every week, Invector Labs publishes a newsletter that covers the most recent developments in AI research and technology. You can find this week’s issue below. You can sign up for it below. Please do so, our guys worked really hard on this:
AI Research
SqueezeBERT
Former Tesla AI Leas Forrest Iandola and researchers from UC Berkeley published a paper proposing a new method for optimizing large language models to run un mobile devices.
Self-Supervised Video Understanding
AI researchers from Facebook published a paper outlining a self-supervised model for video analysis.
>Read more in this blog post from the Facebook AI Research team
Cross Language Skills Benchmark
Microsoft Research published a paper proposing XGLUE, a new benchmark dataset for evaluating cross-lingual transfer capabilities in natural language models.
>Read more in this blog post from Microsoft Research
Cool AI Tech Releases
NetHack Learning Environment
Facebook Research released a new reinforcelement learning environment in the famous NetHack game.
>Read more about it in this blog post from Facebook Research
TensorFlow Privacy Library
TensorFlow released a new library that enables the testing of privacy features in deep learning models.
>Read more in this blog post from the TensorFlow team
New ML Kit
Google announced some changes in its on-device machine learning framework, ML Kit, including its decoupling from Firebase.
>Read more in this blog post from the ML Kit team
AI Startups and Relevant News
NLP Startup Raises $26 Million
RASA, a startup building a platform for conversational AI agents, raised $26 million from prominent venture investors such as Andreesen Horowitz.
>Read more in this coverage from TechCrunch
New Apple’s Core ML
Apple’s mobile toolkit got a lot of new and exciting updates including the ability to update models on the fly.
>Read more in this coverage from VentureBeat
Apple’s AI Strategy
Apple is executing in an AI strategy focused on delivering small conveniences to users instead of large AI initiatives.
