Title: t-SNE Reduction Method in Star Spectral Classification

Speaker: Ma Xingyun（PMO）

Time: 12:15pm, November 7, 2025

Location: 6-215, PMO Xianlin Campus

Abstract: Spectral analysis is a key method for understanding the physical and chemical properties of stars. With modern observational technology, tens of millions of stellar spectra are now available. Analyzing this vast amount of spectral data more efficiently and rapidly presents a new challenge. Concurrently, the rapid development of computer technology has produced a plethora of algorithms. Many studies have already attempted to integrate these algorithms into astronomical research, with t-Distributed Stochastic Neighbor Embedding (t-SNE) being a prominent example. This presentation will explain the significance of spectral classification in stellar studies, the operational principles of t-SNE, and will review two papers that utilize t-SNE for spectral classification, thereby highlighting the method's potential in both classifying spectra and screening for erroneous data.