Interactive human decision-making and knowledge discovery: assistance from software agents with augmented artificial intelligence

The project will explore practical architectures for augmented AI enabled software agents. In particular, he will explore agent architectures that incorporates machine learning, ontologies and Bayesian decision networks for interactive decision-making and knowledge discovery in real world applications.
While ontologies are ideal for representing and structuring domain knowledge, Bayesian decision networks (BDN) provide explicit support for reasoning with uncertainty, and for capturing and reasoning about decision-making processes. Moreover, BDNs capture the decision-making context and systems that incorporate both ontologies and BDNs can potentially enable not only semantic interoperability but also pragmatic interoperability.

Going beyond decision support the agent must also support novel knowledge discovery. Emerging deep neural network techniques like spatial temporal graph neural networks can rapidly learn complex prediction and pattern detection models from real-time sensor observations and can adapt to changes in the agent’s environment. He will take an application use case approach and will focus on two applications that he has worked on previously, i.e. residential household electricity consumption behaviour and investment decision making in stock markets, as examples of complex and erratic socio-physical application environments. The research will provide insights into the design of the next generation of human centred AI systems.

Key Publications

Wanyana, T. and Moodley, D., 2021. An Agent Architecture for Knowledge Discovery and Evolution. In German Conference on Artificial Intelligence (Künstliche Intelligenz) 2021 Sep 27 (pp. 241-256). Springer, Cham.

Toussaint, W. and Moodley, D., 2020. Clustering Residential Electricity Consumption Data to Create Archetypes that Capture Household Behaviour in South Africa. South African Computer Journal, vol. 32(2), pp.1-34.

Pillay, K. and Moodley, D., 2021. Exploring Graph Neural Networks for Stock Market Prediction on the JSE. In Southern African Conference for Artificial Intelligence Research (pp. 95-110). Springer, Cham.