10(ish) questions with new faculty member Iris Tien

GT CEE

 

 

September 30, 2014

GT CEE Tien

Iris Tien is the newest member of the School of Civil and Environmental Engineering faculty, joining the school this fall after completing her Ph.D. at the University of California, Berkeley. She took a few minutes recently to talk about her work and why it’s important to her.

Q: I was reading a little bit about your background and I saw you started out in medical research when you were an undergrad studying civil engineering. How did that happen?

A: I’m not your most traditional civil engineering person. It started from this idea of sensor monitoring, wireless sensor networks, and maybe expanding the definition of health monitoring beyond just structural health monitoring to look at [human] health monitoring. So this opportunity came up to work with the UCSF [University of California, San Francisco] medical school and their Parkinson’s disease research center to use sensors to see if we could diagnose Parkinson’s disease based on how people walked. That was really interesting to me. I liked working with people, and it seemed a little bit different but potentially fun.

Q: And did that plug into your civil engineering experience?

A: Yeah, absolutely. The use of the sensor, developing the system of collecting data, and the data-analysis part of it is very solidly within engineering and signal processing, pattern recognition, machine learning. That actually informs how I look at data now from structures, doing anomaly detection or classifying data into healthy or unhealthy states.

Q: It’s the same idea, when you put it that way. If somebody’s gait has changed or they have these markers for Parkinson’s, it’s an anomaly. And if this road or bridge or piece of civil infrastructure has a problem …

A: If it’s degrading over time, there’s some sort of anomaly that happens. There’s definitely a parallel.

Q: So you were still thinking you wanted to be an engineer.

A: I definitely wanted to be an engineer. But even now I think I’m open to revisiting that type of research. Especially here at Georgia Tech with the Emory connection and the CDC [the U.S. Centers for Disease Control and Prevention], I think there’s a lot of potential opportunities. The opportunity to work with people and to have that impact on people’s lives is pretty cool.

Q: We talk a lot of about the interdisciplinary research, and the idea that you can take these engineering tools and use them in health probably applies in other areas too.

A: Absolutely. The two areas that I’ve thought about are health and business. I look at dynamically evolving systems, so you have a human as a dynamically evolving system over time, you have structures that degrade over time, you have humans whose health might degrade over time that you can monitor, and you have businesses that have constantly evolving information.

I like to work at the boundaries between things and cross across disciplines and fields to take ideas or techniques that have been developed in one field and apply and develop them to solve new problems in different fields.

Q: In your research you use Bayesian networks that assess the reliability of infrastructure systems. Tell me about those networks.

A: You have individual components of a system and each of those components is connected. A Bayesian network models each of those individual components and the connections between them. What makes them special or powerful is that each of those components are probabilistic: there’s a probability that a component will be in a certain state — to use a simple example, “healthy” or “unhealthy,” or “failed” or “survived.” There’s some likelihood that a component is in one of these states. And then when you’re looking at the connections between the different components, those are also probabilistic. So if this component is dependent on this other component, that relationship is also modeled in a Bayesian network.

Q: What do these models tell us about the reliability of a system or a structure?

A: One of things that you can do is identify the critical points of a structure and see the effect of one individual component on the overall system. With Bayesian networks, when you have new information about one part of the network, you can input that one part into the network and then that information propagates throughout the entire network. It will update all of the states of the other components in the system based on what new information you have put in. So when you’re thinking about what does it tell you about a system, you can identify the critical parts of a system, you can say these are the components or the parts that are the greatest vulnerability of the system. You can also look at the effect of different management or rehabilitation actions on the system. You can say, what if I make this particular part stronger? How would that affect the other components? How would that affect my overall system performance?

Q: Why do you do this work? Why is it important to you?

A: Part of what drew me into civil engineering in the beginning was that I like to work with things that you can see around you and that people use every day. These are the systems that people come into contact with every day and that people rely on — when you turn on your faucet, water is going to come out; when you switch on a light, the light is going to turn on; you can get from point A to point B in some sort of reliable manner. That makes the work important and makes it worthwhile. You’re trying to improve these systems and improve the reliability and trying to minimize the risk of them failing over time. And on the more theoretical side, it’s kind of like solving a puzzle. How do all these systems fit together and how do you model them? I’ve always really liked solving puzzles.

Q: We’re talking about wear and tear, or the life cycle of infrastructure, but there’s a disaster or unexpected-event aspect to your work, too.

A: Exactly. That’s the hazard analysis part of it. With a Bayesian network, you can model a hazard. Say a hurricane blows through. What is the effect on these individual components, and what are the effects of these individual components on the overall system? Those connections are something that I’m working on right now: what’s the best way to model those relationships and have those interdependencies between a hazard and a system and the interdependencies between systems? You can run scenarios as well. If you have this type of earthquake happen, what is the effect if you retrofit this part of the system? And then, if another earthquake happens, what’s the effect? If, after an earthquake, you can bring this part of the system up quicker, how does that affect the overall system?

Q: How does this tool actually get used?

A: The whole idea is to support decision-making and help decision makers better design, manage, rehabilitate, repair, or even do preventative things to help make these systems more reliable. One of the benefits of a Bayesian network is that it’s a pretty visual thing. You can model out your system and you can look at it. Once the underlying structure is there, it’s almost ready-made to be used by someone who maybe doesn’t have a Ph.D. in engineering but who has a deep understanding of that system and how it works.

Q: This is your first faculty position. How’s it going so far?

A: On the teaching side, I really like interacting with the students. [On the research side,] Being at Tech and having all these people around, I’ve had lunches with a bunch of people and met with people to talk about writing proposals together and have submitted a couple of collaborative proposals. That part has been fun, to meet with people and to talk about ideas, and to think about how I want to shape my own research program. It’s a little bit of a whirlwind.

GT campus

Q: Why did you choose to come to Georgia Tech?

A: In the end, I think there were a few things that really drew me to Tech. The location definitely helped. I think being in a big city is great. Having the amenities of being in a big city has been good. Being close to a big airport is convenient for travel. I was thinking about being the go-to person in reliability in the School, and that was something that was attractive to me, to be the expert in reliability and probabilistic methods and probability. Being in a larger department I thought was a plus because any problem I ever thought about working on, both in civil engineering and engineering in general, I could probably find someone at Tech to work with or talk with about that problem. It seemed like a good place where there would be a lot of opportunities throughout my career. And then, at the end of the day, it was just a feeling about where I would be happy.

Q: I’ve also read that you are a pretty big basketball fan.

A: Yes, I like playing basketball a lot.

Q: Have you played since you got here?

A: Unfortunately I have not. I was just thinking about this the other day. I have my basketball shoes sitting there, and I haven’t used them since I got here. I’ve started trying to bike into work once or twice a week, to have that be both my exercise and commute. I live about 5 ½ or 6 miles away, so it’s a decent distance, and Atlanta is hilly enough to make it interesting.

Q: What else do you like to do in your free time?

A: I sing in a choir. It’s a classical choir. We sing a range of songs, but I like singing classical music. We are singing this 16th century motet right now, and it’s kind of out of the Gregorian tradition. I really like that sort of thing.

Posted by on September 30, 2014 in Media

Leave a Reply