Artificial Intelligence Powered Multi-Dimensional Feature Analysis of Osmo-Priming on Pepper Seeds

Abstract Priming is a commonly used method to increase seed germination performance. In literature, it is seen that the studies investigating the effects of chosen priming parameters on seed germination are limited to one or two features, in other words low-dimensional. Moving to high dimensional space increases the complexity to be analyzed exponentially. Artificial intelligence is known as an important tool in solving multi-dimensional semantic relations and increasing estimation performance. However, artificial intelligence applications in the field are mainly carried out through image processing and do not target the seed germination performance metrics. In this study, as a novel approach, the effect of osmo-priming applications with different input parameters is investigated directly on metrics that are frequently used in the field to define germination quality in a multidimensional fashion using both classical and artificial intelligence methods. It is seen that the difference between varieties had a great effect on other metrics affecting germination performance, therefore, modeling based on varieties could be developed. In addition, unexpected parameter sets are observed that create instability in the close germination trend, therefore, generally used linear estimates may be improved with the substitutions of high-dimentional approaches. Furthermore, the most important attributes and attribute importance order vary among varieties and some varieties are less sensitive to the selected parameters, i.e. more stable, than others.