报告主题：大小数据科学中的计算方法
报告人：崔春风 博士后 （美国加州大学圣芭芭拉分校）
报告时间：2019年6月2日（周日）9:30
报告地点：校本部G508
邀请人：徐 姿
报告摘要：Due to the wide deployment of social networks and mobile devices, massive data is generated on the internet every second. Such big-data resources open new opportunities for artificial intelligence and machine learning, but meanwhile they also cause new challenges in data storage, representation, and computation. On the other hand, in many engineering designs (such as robotic systems, antennas, electronic and photonic ICs), obtaining data samples by measurement or numerical simulation is expensive, therefore people have to verify and optimize complex engineering designs with limited small data sets.This talk will present some computational techniques to address the challenges in both small- and big-data problems. Specifically, I will present some algorithmic solutions by exploring the interface of tensor computation, uncertainty quantification, and machine learning. Firstly, I will present a global optimization method for solving the non-convex tensor eigenvalue problem, which has been applied to the hypergraph matching and latent variable modeling. Then, I will describe a new uncertainty quantification framework with theoretical performance guarantees even if the input uncertainties are non-Gaussian correlated. Our method allows us to efficiently predict and analyze the performance uncertainties of multi-domain systems (e.g., electronic and photonic IC, autonomous systems) with a small number of simulation data samples. Finally, motivated by the implementation of deep neural networks on edge devices, I will talk about a structural analysis framework of the deep neural network, with a goal to reveal how many neurons and layers are necessary.
欢迎教师、学生参加！