EMITLab in the Center for Assured and Scalable Data Engineering (CASCADE) at Arizona State University carries out research on data-, machine learning-, and AI-systems that couple real-world data, simulations, and causal models to support informed and effective decision making in critical complex dynamic systems, such as disasters, pandemics, water resources, and energy infrastructure. Grounded in NSF and USACE-funded work, including DataStorm, PanCommunity, PANAX, pCAR, CausalBench, Designing Nature to Enhance Resilience for Built Infrastructure in Western U.S. Landscape , PIRE: Building Decarbonization via AI-empowered District Heat Pump System, and the NSF Center for the Analysis of the Epidemic Expansion, the lab turns sparse, noisy, and complex data and models into actionable insights for robust and resilient human-centric environments.
Research projects, especially those funded by NSF and USACE, anchor EMITLab’s work in urgent settings (disaster planning, epidemic response, sustainability, and smart buildings) while driving fundamental advances in data management and analysis for trustable decision making.
Infrastructures for exploring “alternative timelines” of disasters.
DataStorm develops data and computation fabrics that couple large‑scale simulations with real‑world data streams to enable disaster planning and response. EMITLab uses these platforms to help planners test hazard scenarios, compare strategies, and understand where small changes in behavior or infrastructure lead to large differences in outcomes.
Multi‑model platforms for pandemic prediction and response.
These projects integrate specialized epidemic, mobility, and behavior models with data sources to explore how interventions impact communities. EMITLab develops the data and model driven simulaiton and machine learning techniques, enabling decision‑makers to weigh health, economic, and social trade‑offs under uncertainty.
Data science for portfolios of natural & built water infrastructure.
In collaboration with the Network for Engineering with Nature (N-EWN), the "Designing Nature to Enhance Resilience for Built Infrastructure in Western U.S. Landscape" project develops spatio‑temporal models and causal tools to evaluate how wetlands, dams, recharge basins, and other interventions affect flows, ecosystems, and communities. The project team further develops reinforcement‑learning‑driven causal discovery and hydrologic forecasting.
Data platforms and AI for efficient, low‑carbon building systems.
From energy‑simulation data management to AI‑empowered district heat pump systems, these efforts use data engineering and learning to detect HVAC faults, optimize operation, and support human‑centered, decarbonized districts. EMITLab’s models must work across testbeds, data regimes, and operating conditions.
Plausible causal discovery and reproducible evaluation.
The pCAR project investigates how to uncover plausible causal structure in noisy, dynamic data, while CausalBench focuses on rigorous benchmarks and services for testing causal algorithms. Together, they ground EMITLab’s work on spatio-temporal forecasting, infrastructure planning, and decision platforms that leverage causal reasoning.
Saliency‑aware representations for noisy, multiscale sequences.
EMITLab develops models and architectures that highlight salient subsequences in multivariate time‑series and fuse raw and salient views via cross‑attention. This research is supported by a constellation of NSF awards on epidemic simulations, building data management, big data, causal learning, and building services, including Data Management for Real‑Time Data Driven Epidemic Spread Simulations , E‑SDMS , DataStorm , BIGDATA: Discovering Context‑Sensitive Impact in Complex Systems , PFI‑BDMC , and pCAR .
EMITLab’s mission is to design data and AI systems that make complex physical and social systems understandable, actionable, and resilient. Rather than assuming perfect data and unlimited resources, the lab embraces sparcity, noise, heterogeneity, and uncertainty and builds platforms that support critical decisions in disasters, pandemics, water management, and energy systems.
EMITLab envisions computing infrastructures for data, models, and causal inference that are as dependable. In this vision, planners and decsion makers can explore “what‑if” futures through interactive simulations/digital twins, assess nature‑based interventions with causal evidence, and operate buildings and districts that are both comfortable and efficient.
Current students advised by EMITLab faculty, including MS, McS, and PhD theses.
Past students advised by EMITLab faculty, including MS, McS, and PhD theses. Public thesis links are provided where available via ASU KEEP, CORE, or other repositories.