This week in AI & Machine Learning: Google’s next gen AI chip, Facebook unsupervised speech recognition, pizza robots, TensorFlow Lite for Microcontrollers, and more!
🤖 Artificial Intelligence News:
- Google Launches the next generation of custom AI chips ← Another huge advancement in computing chips for artificial intelligence! Google had a lot to announce this week at Google IO 2021. You can read a list of the biggest announcements here.
- Facebook AI cuts by more than half the error rate of unsupervised speech recognition
- The Navy sub commanded by artificial intelligence ← We’ve seen a lot of AI examples piloting aircraft, but not many controlling large submarines!
- Why federated learning is the right solution for healthcare AI ← Federated learning seems like it could be the future of data for healthcare applications! Openmined has been doing a lot of work in this field. I welcome you to check them out.
- Artificial Intelligence: How Non-Tech Firms Can Benefit
- Slicing up Pizza Robots
🛠️ Developer Tools & Education:
- Hyperparameter tuning with scikit-learn and Python
- TensorFlow Lite for Microcontrollers ← I’m excited to get these new examples on a raspberry pi pico!
- High Fidelity Pose Tracking with MediaPipe BlazePose and TensorFlow.js
- Building with the Responsible AI Toolkit ← A useful new toolkit from google to explore your datasets.
- TensorFlow Forum ← TensorFlow now officially has a forum!
🎤 Interesting Podcasts & Interviews:
- Sam Harris: Consciousness, Free Will, Psychedelics, AI, UFOs, and Meaning
- AI Researcher James Kahn Explains Deep Learning’s Collision Course with Particle Physics
- Using AI to Map the Human Immune System with Jabran Zahid
- Learning Long-Time Dependencies with RNNs w/ Konstantin Rusch
- Peter Chen on building brains for robots in the real world
- Apache TVM and OctoML
📄 Notable Research Papers:
- Differentiable SLAM-net: Learning Particle SLAM for Visual Navigation
- Style-Restricted GAN: Multi-Modal Translation with Style Restriction Using Generative Adversarial Networks
- Self-Supervised Learning for Fine-Grained Visual Categorization
- Rethinking the Design Principles of Robust Vision Transformer
- Divide and Contrast: Self-supervised Learning from Uncurated Data
- Omnimatte: Associating Objects and Their Effects in Video