Final Report Summary - BAYES-KNOWLEDGE (Effective Bayesian Modelling with Knowledge before Data) The project has helped to improve risk assessment and decision-making in a wide range of critical applications (including medicine, law and forensics) by developing what we call a 'smart data' approach. In contrast to the current popular 'big data' approach - which assumes that applying sophisticated machine learning algorithms to ‘big data’ sets will automatically find solutions to complex risk problems - the 'smart data' approach combines expert judgement (including understanding of underlying causal mechanisms) with relevant data. The approach uses Bayesian networks (BNs) which provide workable models for combining human and artificial sources of intelligence even when big data approaches to risk assessment are not possible. BNs describe networks of causes and effects, using a graphical framework that provides rigorous quantification of risks and clear communication of results. While BNs are not new, what the project has produced is a suite of new methods and algorithms (implemented in a toolkit) that overcome many of the known impediments to their use by practitioners. This includes novel methods for easily constructing useful BN solutions for common risk problems, drastic improvements to accuracy of BNs that involve numeric variables, and novel support for easily incorporating expert judgement with data into the BNs.A nice published summary of the project can also be found in: Fenton, N E (2018) "Evidence based decision making turns knowledge into power", EU Research 'Beyond the Horizon' Magazine, Spring 2018, pp 38-39. PDF version: www.eecs.qmul.ac.uk/~norman/papers/BAYES-KNOWLEDGE_EUR15_H_Res.pdf Another forthcoming summary can be found here: Fenton, N. E., & Neil, M. (2018). "How Bayesian Networks are pioneeringthe ‘smart data’ revolution", Open Access Government, to appear July 2018. PDF: www.eecs.qmul.ac.uk/~norman/papers/OAG19_FENTON_FINAL.pdf
