Online Event | Ph.D. Dissertation Defense | Towards Structured Prediction in Bioinformatics with Deep Learning

Jul 27 2020 04:00 PM - Jul 27 2020 06:00 PM


Using machine learning, especially deep learning, to facilitate biological research is a fascinating research direction. However, in addition to the standard classification or regression problems, whose outputs are simple vectors or scalars, in bioinformatics, we often need to predict more complex structured targets, such as 2D images and 3D molecular structures. The above complex prediction tasks are referred to as structured prediction. Due to the properties of those structured prediction problems, such as having problem-specific constraints and dependency within the labeling space, the straightforward application of existing deep learning models on the problems can lead to unsatisfactory results. In this talk, we argue that the following two ideas can help resolve a wide range of structured prediction problems in bioinformatics. Firstly, we can combine deep learning with other classic algorithms, such as probabilistic graphical models, which explicitly model the problem structure. Secondly, we can design and train problem-specific deep learning architectures or methods by considering the structured labeling space and problem constraints. In the talk, I will first give an overview of my works during my Ph. D. study related to solving the structured prediction problems in bioinformatics. Then, we go into two projects in detail. In the first one, we proposed deep learning guarded Bayesian inference framework for reconstructing super-resolved structure images from the super-resolved fluorescence microscopy data. This framework enables us to observe the overall biomolecular structures in living cells with super-resolution in almost real-time. In the second one, we zoom in on a particular biomolecule, RNA, predicting its secondary structure. For this one of the oldest problems in bioinformatics, we proposed an unrolled deep learning method that can bring us with 20% performance improvement, regarding the F1 score. Further extension of our works to other challenging but important problems, such as healthcare problems, can potentially directly benefit people’s health and wellness.

Brief Biography:

Yu Li is a Ph.D. student at KAUST in Saudi Arabia, majoring in Computer Science, under the supervision of Prof. Xin Gao. He is a member of the Computational Bioscience Research Center (CBRC) at KAUST. His main research interest is developing novel and new machine learning methods, mainly deep learning methods, for solving the computational problems in biology and understanding the principles behind the bio-world. He obtained MS degree in CS from KAUST in 2016. Before that, he got a Bachelor degree in Biosciences from the University of Science and Technology of China (USTC).