Combining data-driven model with domain knowledge: Machine learning approaches for systems biology
主 题: Combining data-driven model with domain knowledge: Machine learning approaches for systems biology
报告人: Dr. Ke Yuan (Cambridge University )
时 间: 2014-02-27 14:00-15:00
地 点: 理科1号楼1303室（统计中心活动）
The ability to generalize practical problems to standard tasks such as regression and classification has made data-driven modelling immensely popular across many scientific disciplines. However, data-driven approaches quite often use little knowledge about the application, making them difficult to interpret. In this talk, we discuss three stories of finding a middle ground between data-driven models and knowledge-driven models. The first story is decoding neural spike trains with a state-space model with point process observations. The second part, we present a model to capture pathway rewiring events from time course RNA inference screens. Lastly, we combine classical phylogenetic methods with Bayesian nonparametric models to characterising intratumor heterogeneity in cancer.
About the speaker（报告人介绍）：Dr. Ke Yuan is a postdoctoral research fellow at the Cancer Research UK Cambridge institute, University of Cambridge, UK. He obtained MSc and PhD both in Electronics and Electrical Engineering from University of Southampton, UK. His doctoral work focuses on Bayesian inference in time series models. Before coming to Southampton, he got his BEng in Telecommunication Engineering from Nanjing University of Posts and Telecommunications. Currently, his research interest is characterising intratumor heterogeneity with machine learning approaches.