|
Research Experiences
Graph Representation Learning for Time Series Forecasting
Research Assistant at Machine Learning and Robotics Group, Advisor: Prof. Xiao-Jun Zeng
University of Manchester, Manchester, UK, Oct. 2020 - Present
Proposed weighting scheme for time series encoding by mapping binary relations into weighted networks, which overcomes information loss of graph mapping. Extended single-layer graph to multiplex networks, and introduced intra-layer and inter-layer similarities to extracts past and recent sequential knowledge. Introduced ProbAttention mechanism to evaluate probabilistic attention distribution for multi-variate long-term sequence modeling and forecasting.
Decision Making Under Uncertainty in Complex Systems
Research Intern at Center for Mathematics and Systems Science, Advisor: Prof. Yong Deng
University of Electronic Science and Technology of China, Chengdu, CN, Jun. 2019 - Sep. 2019
Addressed inner and outer dependence in evidence theory under uncertainty, and achieved higher sensitivity and accuracy in transportation project selection. Developed a decision-making framework applicable to complex systems, applying to products and process strategies selection for produced water management.
Time Series Fusion and Prediction by Complex Networks
Research Assistant at Information Fusion and Intelligent System Lab, Advisor: Prof. Yong Deng and Prof. Fuyuan Xiao
Southwest University, Chongqing, CN, Mar. 2018 - Jun. 2020
Investigate time series data via complex networks analysis, exploring potential links in temporal networks by link prediction based on random walk. Proposed network-based models for time series aggregation and improved predictive accuracy on real-world datasets, offering a novel promising avenue of network forecasting.
|