The Bonsai platform
The Bonsai platform simplifies machine teaching with deep reinforcement learning so you can train and deploy smarter autonomous systems:
Infographic of the Bonsai platform. Shows a simulator icon flowing into a circular graph that connects the training engine, training curriculum, and integrated simulation. Another arrow flows away from the circular graph to indicate an exported brain.
- Integrate training simulations that implement real-world problems and provide realistic feedback during training.
- Train adaptive brains with intuitive goals and learning objectives, real-time success assessments, and automatic versioning control.
- Export the optimized brain as a Linux container and deploy it on-premises, in the cloud, or at the edge.
Note
Project Bonsai runs on the Azure stack and adheres to Azure compliance requirements. Project Bonsai does not store/process customer data outside the geo-location that the customer deploys the service instance in. When you use Bonsai, your Azure subscription is charged the following resource costs:
- Azure Container Registry (basic tier) for storing exported brains and uploaded simulators.
- Azure Container Instances for running simulations.
- Azure Storage for storing uploaded simulators as zip files.
- Azure Monitor for storing logs of brain training and assessment.