Semi-supervised Dimensionality Reduction with Adaptive Neighbors for Intellgent Agriculture

In the intelligent agriculture, the popularity of high-dimensional data highlights the importance of dimensionality reduction algorithms. However, Traditional methods struggle with graph quality. The new semi supervised adaptive neighbor dimension reduction (SAN) algorithm addresses this limitation by seamlessly integrating dimension reduction and graph learning within the label propagation framework. SAN simultaneously learns the projection matrix and refines the graph structure to promote mutual enhancement between the two. Extensive experiments conducted on five different datasets have shown that SAN surpasses current state-of-the-art methods and provides a powerful solution for processing complex high-dimensional data in smart agriculture applications.