Deep Learning has emerged as a powerful tool for addressing complex scientific challenges, particularly in the domains of image generation, parameter estimation, and large-scale data analysis within gravitational theory and cosmological research. Meanwhile, extracting information about the universe directly from observational and simulated datasets, without relying on prior assumptions, has become increasingly significant. This presentation provides an overview of the recent advancements made by our research group in applying novel deep learning methodologies to cosmological studies. Specifically, I will introduce the recently proposed Branch-Corrected Denoising Diffusion Model and its application in black hole image generation. Furthermore, I will discuss the reconstruction of the Hubble parameter based on the Ef-KAN model recently developed by our team, along with our research on enhancing cosmological parameter estimation using Long Short-Term Memory Networks.

Biography

王接词，湖南师范大学教授、博士生导师。主要研究方向是引力和相对论性量子信息理论，发表论文100余篇，单篇引用600余次。主持了国家自然科学基金优秀青年科学基金项目、面上项目，湖南省杰出青年科学基金项目等科研项目。入选了湖南省“科技创新领军人才”、“湖湘青年英才”和 “121创新人才培养工程”等人才计划，获“国家优秀自费留学生”等奖励。担任湖南省物理学会副秘书长和《Sci. China-Phys. Mech. & Astron.》青年编委。

Inviter: Shao-Jiang Wang