Predicting Grass Growth for Sustainable Dairy Farming: A CBR System Using Bayesian Case-Exclusion and Post-Hoc, Personalized Explanation-by-Example (XAI)

By Eoin M. Kenny, E. Ruelle, A. Geoghegan, L. Shalloo, Micheál O'Leary, M. O'Donovan, Mark T. Keane. 2019

24 citations recorded by the source corpus.

Smart agriculture has emerged as a rich application domain for AI-driven decision support systems (DSS) that support sustainable and responsible agriculture, by improving resource-utilization through better on-farm, management decisions. However, smart agriculture’s promise is often challenged by the high barriers to user adoption. This paper develops a case-based reasoning (CBR) system called PBI-CBR to predict grass growth for dairy farmers, that combines predictive accuracy and explanation capabilities designed to improve user adoption. The system provides post-hoc, personalized explanation-by-example for its predictions, by using explanatory cases from the same farm or county. A key novelty of PBI-CBR is its use of Bayesian methods for case exclusion in this regression domain. Experiments report the tradeoff that occurs between predictive accuracy and explanatory adequacy for different parametric variants of PBI-CBR, and how updating Bayesian priors each year reduces error.

Subfields: Artificial Intelligence, Plant Science