This week in AI & Machine Learning: Drones that land like a bird, robots doing brain surgery, 7-Eleven autonomous delivery, building AI-powered apps faster with PyTorch Live, and more!
Authors note:
I want to give a shout out to The Next Byte podcast providing a great overview of computer vision applications and featuring Plainsight’s article on detecting unwanted & hazardous objects with vision AI.
My Top AI Highlight:
Grasping Robotic Drone Can Land on a Branch Like a Bird
You may have seen a drone get stuck in a tree, but have you ever seen a drone gracefully land in one? Well, now you can!
Researchers from Stanford University have developed the “stereotyped nature-inspired aerial grasper” (SNAG) grasping mechanism that can allow drones to land and take off in complex environments just like birds.
Watch the video below and read more here.
🤖 Artificial Intelligence News:
- Tesla releases new footage of auto labeling tool for its self-driving effort
- DeepMind claims AI has aided new discoveries and insights in mathematics
- The Quest to Make an AI that Can Play Competitive Pokemon
- Artificial intelligence that understands object relationships
- Robots doing brain surgery: CEO describes the latest in medical AI
- Better Machine Learning Demands Better Data Labeling
- Twitch will use machine learning to detect people evading bans
- AWS introduces new Trn1 chips to speed up training of machine learning models
- 7-Eleven to pilot delivery via robotics-powered vehicles
- Amazon launches AWS RoboRunner to support robotics apps
- Robotics-Powered Stellar Pizza Plans LA Restaurant
🛠️ Developer Tools & Education:
- Build your AI-powered mobile apps in minutes with PyTorch Live
- NeuralProphet: The neural evolution of Meta’s Prophet
- Continuous Adaptation for Machine Learning System to Data Changes
- RLDS: An Ecosystem to Generate, Share, and Use Datasets in Reinforcement Learning
- Composing Decision Forest and Neural Network models
📅 Upcoming Online AI & Data Science Events:
- DevFest 2021 — AI/ML Edition (Dec 8–9:00 am PST)
- Optimizing Storage Efficiency for Meta Video Processing (Dec 13–12:00 pm PST)
- Maximizing GPU Utilization and Model Accuracy With Data-centric AI Practices (Dec 14–10:00 am PST)
- Practical MLOps: Operationalizing ML Models in Production (Dec 16–10:00 am)
🎤 Interesting Podcasts & Interviews:
- Can AI See Better Than Humans?
- Real or Not Real? Attorney Steven Frank Uses Deep Learning to Authenticate Art
- Multi-modal Deep Learning for Complex Document Understanding with Doug Burdick
- Predictive Maintenance Using Deep Learning and Reliability Engineering with Shayan Mortazavi
- AI-generated code with OpenAI Codex
📄 Notable Research Papers:
- Masked-attention Mask Transformer for Universal Image Segmentation
- HyperStyle: StyleGAN Inversion with HyperNetworks for Real Image Editing
- On the Integration of Self-Attention and Convolution
- End-to-End Referring Video Object Segmentation with Multimodal Transformers
- NeRF in the Dark: High Dynamic Range View Synthesis from Noisy Raw Images
- Object-Aware Cropping for Self-Supervised Learning
🐘 About the Author & Plainsight:
- Sage Elliott is a Developer Evangelist at Plainsight & passionate about making AI more approachable. Connect with Sage on Twitter or LinkedIn.
- Plainsight’s vision AI platform streamlines the end-to-end machine learning process. From data annotation through deployment, customers quickly create and successfully operationalize their own vision AI applications to solve highly diverse business challenges. Join the conversation on Slack.