Apple Grading Based on Evidential Ordinal ELM with Multi-Criteria Decision Making

To improve the automatic grading accuracy of Fuji apples, this paper proposes an apple grading model based on the Evidential Ordinal Extreme Learning Machine, which considers model epistemic uncertainty and incorporates multi-criteria decision making to support both precise and set-valued predictions. To address uncertainty in the classification process, the model incorporates Dempster–Shafer evidence theory and constructs a Sigmoid-based mass function. In addition, the ordered weighted average operator is employed to extend the utility matrix. Three decision criteria are compared: Maximax, Maximin and Pignistic to evaluate their performance in both precise prediction and set-valued prediction for apple grading. Experimental results on a Fuji apple dataset demonstrate that the proposed model achieves a prediction accuracy of 86.36% in the precise prediction task, which is 7.57% higher than that of the traditional ELM model, and that the Maximax criterion performs best in set-valued prediction, demonstrating the effectiveness of the proposed strategy for uncertain apple grading tasks.