This week in AI & Machine Learning: NVIDIA keynote, a record amount of labor robots, better earthquake detection, creating a digital twin of the earth, and more!
My Top AI Highlight:
GTC November 2021 Keynote with NVIDIA CEO Jensen Huang
This week NVIDIA showed off some amazing technology at their annual GTC (GPU Technology Conference), including their own 3D simulation tool to help create synthetic datasets and build digital twins.
“NVIDIA Omniverse™ is a scalable, multi-GPU real-time reference development platform for 3D simulation and design collaboration, and based on Pixar’s Universal Scene Description and NVIDIA RTX™ technology.”
Omniverse was just one of the many AI tools Jensen Huang presented. Watch the full keynote below:
🤖 Artificial Intelligence News:
- Companies Order Record Number of Robots Amid Labor Shortage
- Machine learning refines earthquake detection capabilities
- Using Machine Learning To Make Better Medical Decisions
- Blackshark Raises Money to Generate a Digital Twin of Earth
🛠️ Developer Tools & Education:
- Model Ensembles Are Faster Than You Think
- An Open Source Vibrotactile Haptics Platform for On-Body Applications
- Making Better Future Predictions by Watching Unlabeled Videos
- The shift to generalized AI to better identify violating content
- The first-ever multilingual model to win WMT, beating out bilingual models
🎤 Interesting Podcasts & Interviews:
- Neal Stephenson: Sci-Fi, Space, Aliens, AI, VR & the Future of Humanity
- Metaspectral’s Migel Tissera on AI-Based Data Management
- The Benefit of Bottlenecks in Evolving Artificial Intelligence with David Ha
- Facebook Abandons Facial Recognition. Should Everyone Else Follow Suit? With Luke Stark
- Analyzing the 2021 AI Index Report
- Q&A with machine translation pioneer: The future of MT is multilingual
📄 Notable Research Papers:
- Are Transformers More Robust Than CNNs?
- Data Augmentation Can Improve Robustness
- Masked Autoencoders Are Scalable Vision Learners
- Full-Body Visual Self-Modeling of Robot Morphologies
- The Emergence of Objectness: Learning Zero-Shot Segmentation from Videos
- Kalman Filtering with Adversarial Corruptions
- RAVE: A variational autoencoder for fast and high-quality neural audio synthesis