Getting to Know: Professor Mahdi Imani

Please describe in simple terms the research that you conduct.

My research aims to develop machine learning tools capable of faster, more accurate, and reliable decision making. 

 

What is the significance of your research in terms of practical applications?

The significance of my research is enabling scalable and efficient modeling, learning, and decision making in a wide range of real-world applications. These include accurate diagnosis and prognosis of cancer, which can save many lives and reduce the huge cost of healthcare; or early and accurate identification of cyber-attacks or faults, which can assure the safety of our infrastructure and the environment. 

 

Which of your research areas is most interesting or most challenging to you?  Why?

The most interesting research topic for me is studying genomics and metagenomics systems. These systems are strongly connected to human health through billions of microbes and small molecules that live in and on us. Our associations with microbes are essential for our health through digestion of our food, training of our immune systems, and protecting us from pathogens. My research aims to gain a deep understanding of the fundamental biology of these systems and harness their potential. The biggest challenge, however, is the large scale of these systems, consisting of hundreds of species and trillions of individual interacting bacteria, which makes the modeling, learning, and decision making extremely challenging. 

 

How did you become interested in this field of research?

My interest in this field started from the senior year of my undergraduate studies. As a student in mechanical engineering, I audited a graduate level course “Artificial Intelligence” in the Electrical and Computer Engineering (ECE) department, which made me aware of the fundamentals of machine learning and potential impacts it could have on various aspects of our lives. This persuaded me to continue my master’s and PhD studies in the ECE department, specifically focusing on research topics related to machine learning.

 

What are the challenges or limitations associated with this field of research?

Despite significant progress made in the field of artificial intelligence in recent years, data limitation and large scale complexity and uncertainty in most practical systems prevent the applications of the existing machine learning techniques. For instance, decisions in autonomous cars without knowing their potential risk, or prescribing therapies for treatment of diseases without knowing the potential impacts of therapies could put the system, environment, or human health at risk. Toward this end, there is a need to develop interpretable and efficient tools capable of taking risk into account in learning and decision making processes. 

 

What are the goals of your research?  What are you hoping to achieve or learn?

My primary research is to develop scalable, efficient, and interpretable tools capable of taking unavoidable ethical, economic, and physical constraints into account during learning and decision making processes. These tools will provide urgently needed capabilities for next generation design and discovery in a wide range of domains, including computational biology and cyber-physical systems. Meanwhile, my goal is to provide excellent long-term training opportunities for the next generation of engineers, at the undergraduate and graduate levels, and prepare them for highly interdisciplinary research, through innovative new courses and research training

 

Is there anything about your research that you think makes it unique from other research being conducted in this area?

The unique part of my research is the integration of machine learning tools with Bayesian statistics and control/learning theory to develop mathematical and statistical tools capable of learning and decision making through limited imperfect and noisy available data, and various sources of uncertainty. 

 

What successes or milestones have you reached thus far in your research?

The main achievements of my research thus far include development of multiple machine learning tools capable of large-scale modeling/learning, training my undergraduate and graduate students to get familiar with machine learning concepts and implementing them in an open access software package, and initiating several outreach and connections with several agencies and companies, including the United States Department of Agriculture, IBM, the National Institute of Standards and Technology, Boeing, General Motors, and more. 

 

Are you collaborating with anyone else, and who funds your research?

My research is supported by the National Science Foundation’s Division of Information and Intelligent Systems (IIS) and by the University Facilitating Fund (UFF) at the George Washington University. I am collaborating with multiple professors and researchers around the country, including at the University of California-Irvine, University of Houston, University of Maryland, and the United States Department of Agriculture.