Towards Knowledge Informed Time Series Forecasting

Presented by Dongjin Song, School of Computing, University of Connecticut

Wednesday, March 27 2024
4:00 PM-5:00 PM ET
AUST 163
Webex Meeting Link
Coffee will be served at 3:30 pm in the Noether Lounge (AUST 326)

Time series forecasting has been widely applied across various domains such as healthcare, traffic control, energy management, and finance. Recently, there has been increasing demand to integrate the knowledge of various forms into the deep forecasting models. In this talk, we will delve into the integration of knowledge in the form of structures or pre-trained word token embeddings to bridge the gap between data-driven time series predictions and domain-centric understanding. Specifically, I will first introduce a novel Structural Knowledge Informed Continual Learning (SKI-CL) framework to perform multivariate time series forecasting within a continual learning paradigm, which leverages structural knowledge to steer the forecasting model toward identifying and adapting to different regimes. Next, I will introduce a Semantic Space Informed Prompt learning with a Large Language Model (S^2IP-LLM) to align the pre-trained word token embeddings with time series embeddings and perform time series forecasting based on learned prompts over the joint space. Finally, I will conclude the talk by highlighting challenges and future directions.

Speaker Bio:

Dongjin Song is an assistant professor in the School of Computing at the University of Connecticut since Fall 2020. He was previously a research staff member at NEC Labs America in Princeton, NJ. He received his Ph.D. degree in the ECE Department from the University of California San Diego (UCSD) in 2016. His research interests include machine learning, data science, deep learning, and related applications for time series data analysis and graph representation learning. Papers describing his research have been published at top-tier data science and artificial intelligence conferences, such as NeurIPS, ICML, KDD, ICDM, SDM, AAAI, IJCAI, ICLR, CVPR, ICCV, etc. He is an Associate Editor for Neurocomputing and has served as Senior PC for AAAI, IJCAI, and CIKM. He received the prestigious NSF CAREER award in 2024 and the UConn Research Excellence Research (REP) Award in 2021. He has co-organized the AI for Time Series (AI4TS) Workshop at IJCAI, AAAI, ICDM, SDM, and MiLeTS workshops at KDD.