Taking Machine Learning to the Next Level
summary
Google describes how TensorFlow plus Cloud Machine Learning Engine let teams train scalable models (hundreds of workers, GPUs or TPUs) or call pretrained models via Vision, Speech, Natural Language and Translation APIs. Airbus uses TensorFlow to classify satellite images and reduced errors from 11% to 3%, AXA predicts major claims for pricing, Ocado ranks support tickets with sentiment analysis, and transfer learning via AutoML alpha for vision creates custom classifiers without code while TPUs available in beta accelerate training.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:54How neural networks learn
- 3:42AI use cases at Google
- 4:48TensorFlow and Cloud ML
- 6:36Cloud ML services
- 8:04Pretrained models and APIs
- 9:31Model development challenges
- 10:15AutoML without coding
- 11:58TPU specialized processors
- 12:50Conclusion
related talks
51mCloud AI: How to Get Started Injecting AI Into Your Applications
59mMachine Intelligence at Google Scale
38mReal-world Machine Learning with TensorFlow and Cloud AI
32mMachine Learning for Game Developers
27mKeynote - Google Cloud, Machine Learning
42mMachine Learning APIs, from Training [...]
26mMachine Learning with TensorFlow
07mMachine learning at Google