计算机软件新技术国家重点实验室   

摘 要：

Inspired by the tremendous success of deep generative models on generating continuous data like image and audio, in most recent years, deep graph generative learning is becoming a promising domain which focuses on generating graph-structured data. Most of them are unconditioned generative models which has no control on modes of the graphs being generated. Going beyond that, in this presentation, we will talk about a recent topic named Deep Graph Transformation: given a source graph, we want to infer a target graph based on their underlying global and local transformation mapping. By automatically interpreting such transformation mapping, we aim to discover new rules and patterns of the graph transformation mechanism. Deep graph transformation could be highly desirable in many promising applications on network synthesis, such as chemical reaction simulation, brain network modeling, and protein design and structure prediction. We will introduce our recent progress on new frameworks, graph convolution&deconvolution techniques, and architectures that can handle the graph spectral evolution, in order to fulfill the graph transformation task. Beyond this, we will further introduce the current limitation of optimization techniques for deep learning on complex structured prediction/generation tasks, which further motivate our recent work on gradient-free optimization techniques for deep learning models.

报告人简介:

Dr. Liang Zhao is an assistant professor at the Department of Information Science and Technology at George Mason University. He obtained his PhD degree in 2016 from Computer Science Department at Virginia Tech in the United States. His research interests include data mining, artificial intelligence, and machine learning, with special interests in spatiotemporal data mining, deep learning on graphs, nonconvex optimization, and interpretable machine learning, as well as their applications broadly in life science, cybersecurity, and geo-information systems. He has published over 80 peer-reviewed full research papers mostly in top-tier conferences and journals such as KDD, ICDM, TKDE, Proceedings of the IEEE, TKDD, TSAS, IJCAI, AAAI, WWW, CIKM, SIGSPATIAL, and SDM. He won best paper awards such as Best Paper Award in ICDM 2019 and “Bests in ICDM” at KAIS journal. He was ranked as “Top 20 Rising Star in Data Mining” by Microsoft Search in 2016. He has also won several other awards such as Outstanding Doctoral Student in the Department of Computer Science at Virginia Tech in 2017, NSF CRII Award in 2018, and Jeffress Trust Award in 2019. He has been serving or co-chairing several prestigious venues such as Proceeding Chair of ACM SIGSPATIAL 2020, Sponsor&Exhibits Chair of SecureCom 2020, and Panel Chair of SSTD 2017, co-Chair of GeoAI at SIGSPATIAL 2019, and Co-Chair of DeepSpatial in ICDM 2019. He also regularly serves as TPC/reviewers of top-tier conferences and journals such as KDD, ICDM, ICML, IJCAI, WWW, AAAI, SDM, TKDE, TKDD, and KAIS. His research is funded by several grants from National Science Foundation, as well as grants from other agencies such as Bank of America and Nvidia.

时间：1月13日  10:00-12:00

地点：计算机科学技术楼223室