毎月恒例の水田勉強会。今回の話題は、水稲や水田環境から収集したビッグデータをXAIで解析する手法について。新進気鋭の岐阜大学山口先生による話題提供。ドローンで取得した植物群落の画像からバイオマスや収量などの目的変数だけではなく、LAI、草丈、茎数などの説明変数も推定。説明変数の多重共線性を乗り越えて回帰させAIによりパラメータ推定することで、目的変数の変動要因を明らかにしようとする手法。まさに説明可能なAI。温故知新といわれる作物学と日進月歩の人工知能を融合させた手法。大きなブレークするとなるか。

Monthly regular rice field study meeting. This time’s topic was about analyzing big data collected from rice paddies and paddy field environments using XAI (Explainable AI). The presentation was given by the up-and-coming Professor Yamaguchi from Gifu University. Using drone-captured images of plant communities, the method estimates not only target variables like biomass and yield but also explanatory variables such as LAI (Leaf Area Index), plant height, and tiller number. The approach aims to clarify the factors affecting target variables by overcoming multicollinearity among explanatory variables through AI-based parameter estimation and regression. This is truly explainable AI. It’s a methodology that fuses traditional crop science (described as learning from the past) with rapidly advancing artificial intelligence. Will this lead to a major breakthrough?I’m very excited to see.