Study of Approaches to Preprocessing Datasets and Initializationof Embeddings in Low-Dimensional Space when Implementing the PacMAP Algorithm

The paper considers approaches to solving problems of preprocessing of datasets and initialization of embeddings in lowdimensional space when implementing the PacMAP algorithm. In the basic version of the PacMAP algorithm., the PCA algorithm is used to solve these problems. The proposed study examines the prospects for using 11 dimensionality reduction algorithms (DRA) in PacMAP algorithm in terms of minimizing the value of the loss function (LF) with acceptable time costs. The results of experimental studies have been presented and recommendations on the feasibility of using certain DRAs have been ascertained.