Learns models online while the data flow so that the models are always up-to-date even when the character of the data changes.
Works with both streaming and batch data; can utilize stored big data before feeding directly from the source streams.
Our systems store data temporarily and can scale horizontally to handle large number of features as well as higher stream speeds.
Supports unsupervised, semi-supervised and supervised learning. When semi-supervised, your domain experts are presented only with a minimal set of data for labelling.
Allows rich and easy interactions with the user via responsive web interface.
Works with heterogeneous types of data; text, numbers, time-series, networked-data, features extracted from video etc. all can be fed into out systems.
Employs feature selection algorithms to allow models as simple as possible but not simpler.
Deploys to the cloud as well as on local computing resources; deployment using Docker is possible.
Integrates with existing data streams, schema, and solutions; TCP connections, REST or Websocket connections, database interfaces and file systems are supported.