We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
XAI for Decision Support in Microplastic Pollution Management
Summary
Researchers trained machine learning models on a dataset of microplastic fibers, beads, and fragments, finding that Random Forest achieved the best classification accuracy with particle size and color intensity as the most predictive features, and that integrating SHAP-based explainability reduced expert decision time by over 60% compared to opaque AI outputs.
The continuing increase of microplastic pollution requires urgent development of proper ecological monitoring alongside classification systems. The paper investigates how Explainable Artificial Intelligence (XAI) frameworks react with microplastic pollution management tools to enhance decision processes while increasing transparency capabilities. The training of machine learning models through a dataset consisting of fiber beads and fragments yielded Random Forest as the optimal selection because it provided exceptional accuracy together with ease of interpretation. The key characteristics for prediction turned out to be size, alongside color intensity according to both feature importance and SHAP analysis results. The confusion matrix showed both precise classifications and very low rates of incorrect categorizations. XAI implementation delivered both extraordinary performance outcomes with expert trust enhancement and significantly short time-to-decision of more than sixty percent according to Likert-scale survey results. The examination outcomes establish XAI works to close the distance between machine-generated technical AI outputs and field-oriented implementation decisions. The system provides plain understandable explanations that enable faster and more self-assured action from researchers and policymakers. The study presents an implementation framework for explainable modeling in environmental oversight which serves as a foundation for developing sustainable AI applications in sustainability science.