This a project features web services to train and evaluate neural networks in the cloud using the Kubernetes container orchestration and a Java based microservice architecture.
See ConbexNN-test-scenario.txt
After Login the VM starts autostart.sh
initalizing ConbexNN. The Iris testset and a ViNNSL neural network are automatically imported.
Firefox is preinstalled and opens predefined Tabs
- Kubernetes Dashboard
- VINNSL-NN-UI (user interface showing imported neural networks)
- Swagger API (documentation of the ConbexNN RESTful interface)
- Status Tab (shows status of imported neural networks)
https://kubernetes.host + endpoint
endpoint | Service |
---|---|
/#/ | Vinnsl NN UI |
/vinnsl | Vinnsl Service |
/status | Vinnsl NN Status |
/worker/queue | Worker Queue |
/storage | Storage Service |
/train/overview | DL4J Training UI (while training) |