Machine Learning Orchestrator
mlscript:
- orchestrate streaming data flows for time series analysis based on machine learning models
- feed ensemble machine learning algorithms for supervised learning
- is an interpreted configuration language
- produces declarative infastructure as code
- implemented in one GO binary
- can run as a Jupyter kernel
- builds on and binds to Julia (J), Lua (L), Python (P) and R (R)
- interfaces with Caffe (CA), CNTK (CN), Mixnet (MI), Keras (KE), TensorFlow (TF) and XGBoost (XG)
- supports checkpointing
- implements a Directed Acyclic Graph (DAG)
- scales from a single laptop to large hybrid cloud environments
entities:
- Data: Event (E), Stream (S), MetaData (M), File (F), Cache (C)
- Identity: Credentials (CR), Role (RO), Profile (PR)
- Service: Source (SO), Mapper (MA), Neural Processor (NP), Reducer (RE), Sink (SI)
- Infra: Node (NO), Rack (RA), Availability Zone (AZ), Hosting Center (DC)