Machine learning researcher and scientist of data since the 1980’s. PhD introduced concept of building multiple decision trees to improve knowledge capture from data – ensemble model. Has been an academic at the ANU, a senior researcher at CSIRO, lead data scientist with the ATO and the federal government’s Centre of Excellence in Data Analytics, and now Director of Data Science with Microsoft. Open source advocate for 30 years contributing packages to Emacs, Linux and R and now open source advocate and data scientist with and within Microsoft.
Keynote: Extreme Ensembles as the Future of Data Science and Intelligent Apps
Ensembles have been an underlying and successful concept in machine learning for a long time. The approach continues to deliver significant improvements in modelling, from extreme gradient boosting to deep learning. As we continue to re-learn the importance of privacy and the significance of distributed data and ensembles of models, the concepts come together to bring to data science the concepts of machine learning on the device operating over massively distributed augmented with locally collected data. In this talk I will review the concept and growth of ensembles and propose how we will see the power of ensembles and widely distributed public and private data on the cloud to drive the development of tomorrow’s intelligent applications.