Publications des agents du Cirad

Cirad

Reinforcement learning for crop management support: Review, prospects and challenges

Gautron R., Maillard O.A., Preux P., Corbeels M., Sabbadin R.. 2022. Computers and Electronics in Agriculture, 200 : 14 p..

DOI: 10.1016/j.compag.2022.107182

Reinforcement learning (RL), including multi-armed bandits, is a branch of machine learning that deals with the problem of sequential decision-making in uncertain and unknown environments through learning by practice. While best known for being the core of the artificial intelligence (AI) world's best Go game player, RL has a vast range of potential applications. RL may help to address some of the criticisms leveled against crop management decision support systems (DSS): it is an interactive, geared towards action, contextual tool to evaluate series of crop operations faced with uncertainties. A review of RL use for crop management DSS reveals a limited number of contributions. We profile key prospects for a human-centered, real-world, interactive RL-based system to face tomorrow's agricultural decisions, and theoretical and ongoing practical challenges that may explain its current low uptake. We argue that a joint research effort from the RL and agronomy communities is necessary to explore RL's full potential.

Mots-clés : intelligence artificielle; apprentissage machine; aide à la décision; incertitude statistique; analyse du risque; système d'aide à la décision; algorithme

Documents associés

Article (a-revue à facteur d'impact)

Agents Cirad, auteurs de cette publication :