Staged Online Learner (SOL)

The major objective of this project was to analyse high speed data streams while significantly improving throughput and retaining accuracy levels obtained from the existing state of the art data stream mining toolbox. Currently many organizations are unable to analyse streaming data in real time and are forced to store their data which are later processed in batch mode. The major limitation of this approach is that decisions cannot be taken in timely manner in real time. The project developed of a new framework – Staged Online Learner (SOL) – for data stream mining that explicitly models the level of volatility in the stream and takes advantage of low volatility data segments by re-using models built during counterpart high volatility segments. The project has received the partial 2016 AUT SRIF award.

Project Team:

Russel Pears
Muhammad Asif Naeem

PhD Student:

Chamari Kithulgoda