Multi-source Machine Learning for Time-series Prediction

Vision

We develop deep graph time-series spatio-temporal forecasting models that leverage interdependency structures and contextual information to improve prediction accuracy.

Funded Projects

• Size-segregated Particle Odor Chromatographic Kernel (SPOCK) – IARPA 2023
• A data-driven multi-scale phytotechnology framework for identification and remediation of leached-metals-contaminated soil near coal ash impoundments – DOE 2022
• Robust Real-time Resource Forecasting for Pricing Scheme Design – Adobe 2022
• Time-series Prediction for Cloud Demand Forecasting – Adobe 2020
• Prognostics and System Health Management – Xerox 2015
• An Integrated Approach to Interactive Patient-Centric Monitoring – Xerox 2015
• Baby Data Analytics: Towards Early Detection of Autism – PARC 2014

Publications

• Graph Deep Factors for Probabilistic Time-series Forecasting – ACM Transactions on Knowledge Discovery from Data (TKDD) 2022
• Context Integrated Relational Spatio-Temporal Resource Forecasting – IEEE Big Data 2021
• Graph Deep Factors for Forecasting with Applications to Cloud Resource Allocation – KDD 2021
• Multistage Hybrid Attentive Networks for Knowledge-Driven Stock Movement Prediction – ICONIP 2021 – code, data, presentation
• Predicting Stock Price Movement Using Financial News Sentiment – EANN 2021
• Two-Stage Clustering of Household Electricity Load Shapes for Improved Temporal Pattern Representation – IEEE Access Journal 2021 – code, data, presentation
• Predicting Coordinated Actuated Traffic Signal Change Times using LSTM Neural Networks – Transportation Research Record Journal 2021
• COVID-19 pandemic impacts on traffic system delay, fuel consumption and emissions – International Journal of Transportation Science and Technology (IJTST) 2020
• Estimating Switching Times of Actuated Coordinated Traffic Signals: A Deep Learning Approach – United States Department of Transportation Report 2021
• Assessing the Robustness of LSTM Neural Networks for the Prediction of Actuated-Coordinated Traffic Signal Change Times – Transportation Research Board (TRB) Annual Meeting 2022
• Device health estimation by combining contextual information with sensor data – Patent Granted 2018
• Device Health Estimation by Combining Contextual Control Information with Sensor Data – Workshop on Principles of Diagnosis 2016