Srijita Das

By |

My research is about building sample-efficient machine learning models. My long term goal is to develop collaborative systems that can actively seek advice from humans and make faster decisions, resulting in reliable and practical systems. I specifically focus on design of sequential decision-making models to make them learn faster. We leverage advice from humans in various forms (implicit and explicit) to encourage favorable decisions and avoid decisions having catastrophic consequences. We also focus on minimizing the cost of seeking advice by building suitable machine learning models from historical advice data and reusing them when required. Our research also develops ways to solve complex tasks in Reinforcement Learning by leveraging various kinds of knowledge transfer mechanisms, curriculum learning, teacher-student framework etc. Advances in these directions would make decision-making models sample-efficient and better suited for solving real-world problems. Along the supervised machine learning spectrum, we also focus on problems related to learning with less data, traditionally known as Active Learning, semi-supervised learning, and learning from multiple experts.

Mosharaf Chowdhury

By |

I am a computer scientist and an associate professor at CSE Michigan, where I lead the SymbioticLab (https://symbioticlab.org/). My research improves application performance and system efficiency of AI/ML and Big Data workloads with a recent focus on optimizing energy consumption and data privacy. I lead the ML Energy initiative (https://ml.energy/), a consortium of researchers focusing on understanding, controlling, and reducing AI/ML energy consumption. Over the course of my career, I have worked on a variety of networked and distributed systems. Recent major projects include Infiniswap, the first scalable memory disaggregation solution; Salus, the first software-only GPU sharing system for deep learning; FedScale, a scalable federated learning and analytics platform; and Zeus, the first GPU energy optimizer for AI. In the past, I invented the coflow abstraction for efficient distributed communication, and I am one of the original creators of Apache Spark. Thanks to my excellent collaborators, I have received many individual awards, fellowships, and paper awards from top venues like NSDI, OSDI, ATC, and MICRO.

Zach Landis-Lewis

By |

My research focuses on the use and effectiveness of coaching and appreciation feedback in healthcare. I lead a team that develops a software-based precision feedback system to generate messages about performance to healthcare professionals and teams. My work involves the processing of performance data to detect signals of motivating information that can be delivered with algorithmically prioritized messages, to support performance improvement and sustainment. I lead the DISPLAY-Lab, which collaborates with researchers in a range of clinical and health-related domains, including biomedical informatics, implementation science, and human-centered design.

VAN HAI BUI

Van Hai Bui

By |

Dr. Bui’s research focuses on the operation and control of power and energy systems. We develop energy management systems aimed at optimizing the entire system’s operation to minimize operation costs, enhance system reliability, and improve system resiliency in both normal and emergency operation modes. Recently, the high penetration of distributed energy resources (DERs), including photovoltaics, wind turbines, and controllable distributed generators, in modern power systems has introduced numerous sources of uncertainty, making the operation and control of power systems significantly challenging. Conventional optimization methods often struggle to handle the high uncertainty of DER outputs. With the rapid development of AI/ML algorithms and their wide applications in the engineering domain, these techniques offer potential solutions for operating and controlling power systems. Our research group also investigates the state-of-the-art models in ML, such as deep learning, deep reinforcement learning, and physics-informed graph neural networks, and their applications in power and energy systems.

Peter Reich

By |

Reich conducts global change research on plants, soils, ecosystems and people across a range of scales. His work links fundamental physiology with community dynamics and ecosystem structure and function, from the patch to the globe, within the context of the myriad of global environmental challenges that face us. This includes studying the effects on natural and human ecosystems of rising CO2 and associated climate change, biodiversity loss, and wildfire. This research involves a variety of tools and approaches (long-term experiments, observations, global data compilations, statistical and simulation models), a diverse set of ecosystems (boreal forest, temperate grassland, and more), and a range of scales (local, regional, global). The overarching goal is to understand what we humans are doing to nature in order to help orchestrate a shift towards a nature-forward prioritization that will in turn support and sustain human society.

I studied physics and creative writing and became interested in the fate of our environment; over time I began using tools from each focal area to advance ecological science in a changing world

Johanna Mathieu

Johanna Mathieu

By |

My research focuses on ways to reduce the environmental impact, cost, and inefficiency of electric power systems via new operational and control strategies. I am particularly interested in developing new methods to actively engage distributed flexible resources such as energy storage, electric loads, and distributed renewable resources in power system operation. This is especially important in power systems with high penetrations of intermittent renewable energy resources such as wind and solar. In my work, I use methods from a variety of fields including control systems and optimization. I also use engineering methods to inform energy policy.

Cheng Li

By |

My research focuses on developing advanced numerical models and computational tools to enhance our understanding and prediction capabilities for both terrestrial and extraterrestrial climate systems. By leveraging the power of data science, I aim to unravel the complexities of atmospheric dynamics and climate processes on Earth, as well as on other planets such as Mars, Venus, and Jupiter.

My approach involves the integration of large-scale datasets, including satellite observations and ground-based measurements, with statistical methods and sophisticated machine learning algorithms including vision-based large models. This enables me to extract meaningful insights and improve the accuracy of climate models, which are crucial for weather forecasting, climate change projections, and planetary exploration.

Dani Jones

Dani Jones

By |

Dani Jones’ research program drives CIGLR’s portfolio of research in data science, machine learning, and artificial intelligence, as applied to physical limnology, weather forecasting, water cycle predictions, ecology, and observing system design. This research program aims is to advance societal adaptations to the effects of climate change, including flooding of coasts, rivers, and cities. Dani’s background is in physical oceanography, with specific expertise in adjoint modeling for comprehensive sensitivity analysis and unsupervised classification for data analysis, mostly applied to the North Atlantic and Southern Ocean. In Dani’s current role, they are establishing CIGLR’s new Artificial Intelligence Laboratory, leveraging the institute’s extensive observing assets, datasets, modeling capacity, interdisciplinary expertise, and numerous regional and international partnerships.

Alauddin Ahmed

Alauddin Ahmed

By |

My core research expertise involves developing and employing a wide array of computational methods to discover, design, and characterize materials and systems that address critical challenges in energy and the environment. These methods span from stochastic techniques to molecular dynamics, density functional theory, quantum chemistry, and data science. Beyond contributing fundamental design principles for high-performing materials, my research has led to the discovery of record-breaking materials for hydrogen storage, natural gas storage, and thermal energy storage, alongside creating open-access databases, machine learning models, and Python APIs.

In data science, I have uniquely contributed to feature engineering, compressed sensing, classical machine learning algorithms, symbolic regression, and interpretable ML. My approach to feature engineering involves crafting or identifying a concise set of meaningful features for developing interpretable machine learning models, diverging from traditional data reduction techniques that often disregard the underlying physics. Moreover, I have enabled the use of compressed sensing-based algorithms for developing symbolic regressions for large datasets, utilizing statistical sampling and high-throughput computing. I’ve also integrated symbolic regression and constrained optimization methods for the inverse design of materials/systems to meet specific performance metrics, and I continue to merge machine learning with fundamental physical laws to demystify material stability and instability under industrial conditions.

Looking forward, my ongoing and future projects include employing machine learning for causal inference in healthcare to understand and predict outcomes and integrating AI to conduct comprehensive environmental and social impact analyses of materials/systems via life cycle analysis. Furthermore, I am exploring quantum computing and machine learning to drive innovation and transform vehicle energy systems and manufacturing processes.

 


Accomplishments and Awards

Uduak Inyang-Udoh

Uduak Inyang-Udoh

By |

My research seeks to exploit graph-based modeling theory and the tools of machine learning for efficient control of physical dynamical systems and control co-design in these systems. I am particularly interested in the design of graph-based machine/deep learning model structures that are compatible with basic physics, and using those model structures for real-time actions. Application of interest include advanced manufacturing, thermal and energy storage systems.