I seemed to have entered a new phase of academic life where, instead of applying for postdoc positions myself, I am able to offer them.
Bob Williamson and I have recently secured funding for a research project entitled Structures and Protocols for Inference and we are looking for someone with a machine learning background to join us for 2.5 years as a postdoctoral research fellow. If you think you might be interested read on. If you are sure you are interested, don’t waste any more time and head straight over to the application form.
The text we’ve sent out with the ad summarises the project best:
The project goal is to develop an architecture that will improve the interoperability of machine learning software and make those tools more accessible. This requires a systematic exploration of the types of tasks encountered in machine learning, the development of a language to describe inference tasks, and a theory to relate and compose them.
Essentially, we want to build the foundations of machine learning as web services. While there are already a couple of attempts along these lines—the Google Prediction API being the prominent example—we feel there is room for a broader, more open, more versatile, and consistent approach.
A necessary component of this work will be a solid theory of task specification (structure) and learner interaction (protocols). As well as guiding the design of the ML-as-service architecture we expect such a theory to help clarify and extend some of our existing work that examines relationships between learning tasks such as regret bounds, reductions, and transfer learning.
So, if you have a solid background in machine learning and a penchant for designing software and services, please consider applying for the position.
Mark Reid September 7, 2010 Canberra, Australia