報告題目：Machine Learning in Species Delimitation

報告時間：2025年6月11日（星期三），上午9:30

報告地點：生科院1#109會議室

主辦單位：生命科學學院、江蘇省比較基因組學重點實驗室、江蘇省基因組學國際聯(lián)合研究中心、江蘇師范大學科學技術(shù)研究院

報告人簡介：

　　劉亮，美國佐治亞大學統(tǒng)計系暨生物信息研究所教授。國際分子系統(tǒng)發(fā)育基因組學研究領(lǐng)域新型物種樹方法的創(chuàng)始人之一，曾獲2008年度國際系統(tǒng)生物學家協(xié)會優(yōu)秀科研獎。長期擔任Systematic Biology, Bioinformatics, Journal of Mathematic Biology, Molecular Biology and Evolution, Molecular Ecology 等國際學術(shù)期刊的評委，在Science、PNAS、National Science Review, Systematic Biology、Molecular Biology and Evolution、Bioinformatics等國際學術(shù)期刊發(fā)表論文80余篇，論文總引用次數(shù)約3.5萬余次，單篇論文最高引用2.4萬余次。擔任美國國家自然科學基金委員會二審評委。

報告摘要:

　　Species delimitation is a fundamental task in systematics and biodiversity research, yet it remains a challenging endeavor due to the complex nature of evolutionary processes and the limitations of traditional analytical techniques. Recent advances in machine learning (ML) have significantly enhanced the field of species delimitation by introducing powerful computational tools capable of analyzing complex, high-dimensional datasets. ML algorithms provide a flexible and scalable framework for identifying patterns in genomic and phenotypic data, overcoming many of the limitations associated with traditional taxonomic methods. This paper reviews current applications of ML in species delimitation, highlighting its capacity to resolve issues related to gene flow, incomplete lineage sorting, and cryptic diversity. We present simulation studies that demonstrate the effectiveness of ML approaches and discuss their advantages in improving the accuracy, objectivity, and efficiency of species boundary inference. Our findings underscore the transformative potential of ML in systematics and call for broader integration of these tools in taxonomic research.