EPSRC Grant

Closed Loop Machine Learning

Aims

Machine Learning systems that produce human-comprehensible hypotheses from data are being increasingly used for knowledge discovery within both business and science. These systems are typically open loop, with no direct link between the Machine Learning system and the collection of data. This project will test the alternative of Closed Loop Machine Learning systems. This is related to the area of active learning in which the machine learning system actively selects experiments to discriminate between contending hypotheses. In Closed Loop Machine Learning the system not only selects but also carries out these experiments in the learning domain. In particular this project will test the following two hypotheses.

Hypothesis 1: Closed Loop Machine Learning efficiently converges to accurate hypotheses.

Hypothesis 2: Closed Loop Machine Learning systems can be physically realised using robotics and successfully applied to a discovery task in functional genomics.

To test these hypotheses we will develop the Inductive Logic Programming (ILP) system ASE-Progol (Active Selection of Experiments within Progol). ASE-Progol will use Progol to form the initial hypothesis set. It will then devise experiments to select between competing hypotheses, direct a robot to perform the experiments, and finally analyse the experimental results. ASE-Progol will then revise its hypotheses and repeat the cycle until a unique hypothesis remains. The proposed project is, to our knowledge, the first attempt to use a robot to carry out experiments selected by active learning within a real world application.

Principal Investigator: Prof. S.H. Muggleton

Co-Investigators: Dr R.D. King, Prof. D.B. Kell, and Prof. S.G. Oliver

RA: Dr. C.H.Bryant

Industrial Collaborators: Dr C. Rawlings of Smith-Kline Beecham and Dr W. Blackstock of Glaxo Wellcome.

Dates

Outside Funding