learning framework for self-aware computing
Key Points
- DAIRNet Theme: Distributed Multi-Domain Networks
- University of New South Wales and Consilium Technology
- Point of contact: A/Prof Lina Yao
This project aims to develop a software package for building a durable self-aware software system exhibiting both generalisation and adaptation capabilities. It will output machine-learning models that can quickly generalise and adapt using only a small number of labelled samples to deal with rapidly changing environments and new situations.
The project includes the detection and tracking of trends and other changes in the network and (if available) environmental data and adapting to those changes rapidly. Meanwhile, the output models will exhibit memory, capable of remembering key pieces of evidence and events in the past to reflect and act on cyclical patterns in data over time. The project outcome will enable AI systems to work well on small datasets and be applicable in an ever-changing world for significant periods without manual re-engineering.
