Mission & Vision

The University of Maryland Institute for Advanced Computer Studies, known by its acronym UMIACS (pronounced YOU-me-acks), is a multidisciplinary computing research institute led by distinguished researchers and supported by a cutting-edge infrastructure. UMIACS pioneers computational science involving national defense, precision medicine, big data, cybersecurity, language and culture, and more.

Director's Message to UMIACS Feb '24

I hope you’re enjoying the new year (both the calendar year and Lunar New Year). Given it’s now 2024, I asked someone on staff to investigate our archives for any milestone markers for UMIACS.

We have several documents showing then-University of Maryland President John S. Toll authorizing the formation of UMIACS in Fall 1984. In speaking with others, we determined our institute hit the ground running—with faculty and staff budget lines and significant work being done—in early 1985.

That means 2025 will mark four decades of forward-looking research, innovation and scholarship by UMIACS faculty—relying heavily on assistance from our talented grad students and support staff. We’ll let you know this fall if we plan to have any special recognition of this milestone next year.

But the look back did get me thinking at just how cutting-edge our institute has been from day one. Early pioneers—Rita Colwell in computational biology, Ben Shneiderman and Catherine Plaisant in human-computer interaction, Hanan Samet in spatial data structures, and Jack Minker in AI and deductive databases—conducted groundbreaking research that, for the public, almost seemed like magic.

But we’re not a community of magicians. We’re hardworking scientists and scholars continuing to push forward with bold new ideas. A quick peek at our newsletter will confirm just that.

We continue to be a leader in AI, using a socio-technical approach to improve the trustworthiness and safety of technologies that are transforming society and changing our way of life. We’re exploring the microbiome, using powerful computational tools to investigate the connectivity between microbes interacting with each other, the environment, agricultural systems, and human and animal health. We’re part of a team using data science and quantum sensors to reduce food insecurity and waste. And we’re continuing our trailblazing work in quantum computing, helping design a novel system that has a record number of qubits.

But we’ve never rested on our laurels before—we’re always pushing forward with new scientific discoveries. I expect we will continue that practice, which will require new talent and new ideas.

I’m pleased the newsletter highlights those topics as well, recognizing the superb early-career work by Christopher Metzler and Sanghamitra Dutta. Bravo to every member of our research community! The work you do is important, meaningful to society, and forward-looking. It’s not magic—it’s cutting-edge science.

—Mihai Pop, UMIACS Director

Director's Message to UMIACS March '24

Technology that is part of everyday life. We’ve been “there” for a while now. But the explosion of AI in the past few years offers food for thought on topics like trustworthiness, empowerment, inclusiveness and accountability as they relate to computing and new technologies.

We’re fortunate to have researchers in UMIACS that are addressing these questions—and more—in a meaningful way. From my view as UMIACS director, I see a common thread that is driving much of this activity: creating technology for the greater good and making scientific discoveries that improve people’s lives.

Those ideas are certainly on display in our featured story this month, an engaging video highlighting a multidisciplinary effort to help new mothers answer questions about maternal and infant health care. Jordan Boyd-Graber, one of our top experts on making machine learning more useful, more interpretable, and able to learn and interact from humans, is handling the computational end of the project.

Jordan and the rest of the team are relying heavily on Neha Srikanth, a Ph.D. student in our Computational Linguistics and Information Processing (CLIP) Lab, to manage tech-related tasks like optimizing the AI-powered question and answer software and determining best practices for hosting the platform.

Bravo to Jordan and Neha for their innovative AI work that will have a real impact in on the everyday lives of people in our community.

Nirupam Roy and his students are on that same track, developing AI software called TalkLock that can help identify deepfake videos and stop the spread of disinformation. The project has received a lot of media attention, including this feature from a Baltimore television station.

I also encourage you to read the Q&A with Hal Daumé III in Roadmap magazine. Hal lays out, in a thoughtful and precise way, the importance of ongoing work involving trustworthy AI and why new technologies being developed should help humans, not replace them.

All the projects I just mentioned benefit immensely from the support they receive from UMIACS. As AI research and scholarship continues to expand on this campus and beyond, it will take experienced technical support to ensure that the large computational infrastructures required are built and maintained with the utmost efficiency.

That’s exactly what we do in UMIACS. We have almost 40 years of experience designing, installing and maintaining high-performance platforms for research computing.

I hope you enjoy the upcoming Spring Break and look forward to seeing many of you in person afterward.

—Mihai Pop, UMIACS Director

Director's Message to UMIACS May '24

UMIACS recently hosted a trio of visitors tasked with conducting an external review of our institute. The reviewers came from academic units that somewhat mirror UMIACS—high-level, computing-focused groups at universities that require a team effort from faculty, students and staff to excel.

While a written report from our external reviewers has not yet been delivered, I was offered some anecdotal evidence through brief conversations with the committee members and those they spoke with.

UMIACS is doing very well. From our research endeavors to our computational infrastructure to the quality of our business operations and more, we set the bar high for how a top-tier research enterprise should operate.

This should be of little surprise to many members of our community. We collectively work hard, prodding each other in a focused yet professional way to improve our daily operations.

All the reviewers spoke of how impressed they were by our support staff. I may have even sensed a bit of envy from several of their comments!

Is there room for improvement? There is always room for improvement. That’s the purpose of a comprehensive external review—to highlight the good and identify areas that need attention.

But overall, I believe we came through the review process with flying colors. I cannot say enough how proud I am of the work done by each of you.

Thank you. Enjoy a break to recharge when the semester ends. We’ll see some of you over the summer and the rest of you this fall.

—Mihai Pop, UMIACS Director

Director's Message to UMIACS June '24

While the summer months offer a time to rest and recharge for many of us, the ongoing work in UMIACS continues unabated.

We continue to push forward together—making new discoveries, developing novel technologies, building interdisciplinary partnerships, garnering professional accolades, and welcoming new employees.

These topics, and more, are covered in this month’s newsletter. A common theme is that all our success stories involve teamwork.

Even individual awards and accolades—such as our adjunct faculty member and QuICS Fellow Alexey Gorshkov receiving an IEEE award for his groundbreaking work in quantum electronics—recognizes that success is rarely a solitary effort.

To quote Alexey from the article: “I am profoundly grateful to my numerous fantastic collaborators, including students and postdocs, and to my colleagues—all of these people were instrumental to completing the research that led to this award.”

This idea is echoed by Rachel Rudinger in response to her receiving the prestigious NSF CAREER award. “I am fortunate to work with a group of outstanding Ph.D. students here at the University of Maryland, and it is gratifying to be able to support them through an award that could not have materialized without their hard work.”

And Laxman Dhulipala, who was just honored by the ACM with its prestigious Kanellakis award, is quick to point out that the work he was recognized for would not have happened without his teammates. "It’s an incredible honor to have this collaborative work with Guy [Blelloch] and Julian [Shun] recognized by the ACM. They both have taught me a lot through the years,” he said.

I encourage all of you to reflect on the high level of professional camaraderie and collaboration we have in UMIACS. From pooling resources to purchase computing equipment to sharing a cup of coffee to brainstorm new ideas, the age-old concept of “the whole is greater than the sum of the parts” has worked well in UMIACS for decades.

I expect that we will continue our success—as a team—for years to come.

—Mihai Pop, UMIACS Director

Sponsors

Government Agencies

Corporate Sponsors

Managing Data for Disaster Relief

Ashok Agrawala, Louiqa Raschid, Amol Deshpande, and Dana Nau

The inability of searchers to communicate at the World Trade Center in 2001 and the chaos on the ground after Hurricane Katrina highlighted the need for good ways to share information during and after disasters.

Six weeks after 9/11, Ashok Agrawala—using his expertise in wireless communication and location detection—put together Draco, a rapidly deployable system for helping emergency personnel communicate.

Agrawala’s system allows members of search teams to communicate with each other and with a central commander and to leave virtual tags about problems like broken bones or structural damage.

“The situation was nothing short of chaotic on 9/11. Since then, a lot of people have been developing individual technologies” says Agrawala. His system offers a complete communication package that can be deployed rapidly and is particularly good at tracking individuals’ locations using time-of-flight wireless signals, whether people are indoors or outdoors.

This kind of communication system is valuable not only in disasters but in law enforcement. Agrawala is working with University of Maryland police to test some technologies. He is also working with the Washington-Baltimore High Intensity Drug Trafficking Area (HIDTA) program to develop an information brokering system to expedite communication among different law enforcement agencies. Dick Baer, who works for the Washington-Baltimore HIDTA says, “Ashok’s work has been very effective in achieving integration. He is able to take existing capabilities and put it all together.”

UMIACS researcher Louiqa Raschid, who specializes in integrating information from various sources, was the principal database architect for Sahana, open-source software for disaster management that is now used around the world. “Disaster relief isn’t done by one company solving a problem,” says Raschid, “Information systems tend to fail because of the difficulty of sharing information among all the organizations that are involved.”

Sahana helps match resources with those who need them by bringing together data about victims, volunteers, organizers, emergency requests, and shelters. Because Sahana is opensource, people in different regions can tailor the program to their needs and can integrate Sahana with the software they are already using.

Versions of Sahana have been used in the United States, Pakistan, the Philippines, Indonesia and Sri Lanka. Among its diverse capabilities, the software can send out alerts to large numbers of people. For example, after a recent earthquake, people in Sri Lanka received text messages warning them of possible tsunamis. In 2007, Sahana received the Free Software Foundation’s award for Social Benefit.

“Louiqa is very passionate about using information technology to help solve humanitarian problems,” says Kristina Lerman, a computer scientist at the University of Southern California, who recently wrote a research proposal with Raschid. “Sahana is being used by New York City. FEMA is considering it. It is quite exciting, and Sahana developers are not profiting from it.”

Another researcher in UMIACS, Amol Deshpande works on methods to manage data generated by tiny, wireless sensors using statistical modeling. Whether sensors are deployed for intense, short-term use such as monitoring a wildfire or for long-term, preventative tasks, such as observing a bridge, being able to obtain data in an energy-efficient way makes sensors much more usable, and Deshpande also works to create energy efficient algorithms for sensor networks.

Deshpande’s methods need to cope with the abundant but noisy and incomplete data continuously generated by sensors and must be able to detect anomalies in real time. “Our goal is to make the analysis faster and more efficient,” says Deshpande.

One of Dana Nau’s main research interests is how to generate plans. David Aha, a computer scientist at the Naval Research Lab, says, “Dana is one of the preeminent people in the world of artificial intelligence planning.” In collaboration with Aha, Nau helped develop methods for planning evacuations of noncombatants in emergencies, creating a system that Nau describes as “a sort of intelligent bookkeeper.”

“At certain points, you could ask the system to use our SHO P software to do some of the planning for you,” says Nau. SHOP, for Simple Hierarchical Ordered Planner, can generate plans for tasks in many different kinds of problem domains, provided that a user gives it information about how to do so.

SHO P and its successor SHO P2 are open-source software and have been used in hundreds of projects by government, industry, and academia. A company in Minneapolis called SIFT, for Smart Information Flow Technologies, has adapted SHO P2 to create plans for running unmanned aerial vehicles. A user can enter plain-language instructions for the vehicles, which the program decomposes into computer commands. Robert Goldman, a senior scientist at SIFT, says. “One of the most appealing things about the SHO P system is that it’s adaptable. It doesn’t rely on a very rigid problem structure. To do that, you have to be open to the complicated and messy aspects of real problems.”

Better Systems Software

Jeff Foster and Michael Hicks

Jeff Foster and Michael Hicks are developing ways to help programmers write better systems software—software that is more secure, reliable, and able to run without crashing or needing to be restarted.

Systems software—like the code running workplace servers or Internet shopping sites—must handle many activities in tandem, manage resources like memory and processing power efficiently, and provide ready access to information while also protecting data from unwanted access and tampering. www.umiacs.umd.edu

Hicks and Foster are working to design software that can do all this and also run for years without crashing or needing to be restarted.

Foster and Hicks joined UMIACS within months of each other, both having worked previously on analyzing code for security problems. They continued that work at UMIACS and started collaborating soon after they arrived. In late 2006, they formally merged their research groups. “About half of our students are officially co-advised,” says Hicks, “and the other half are unofficially co-advised.” Everyone in the group meets frequently to go over their projects and problems.

Their research is both formal and mathematical as well as experimental and practical, the two researchers say. “We build programs that relate to existing software today so that we know that the work that we’re doing formally and theoretically will also have practical benefits,” says Foster. One of the problems Hicks and Foster have undertaken is how to update software while it continues to run. For example, an online service or a personal computer could have bugs corrected or get software updates without having to be shut down and restarted. The researchers have designed a flexible and safe system for converting source code into a version that can continue to operate while it is altered. On test runs, Foster and Hicks have shown that common applications, like secure shell servers, FTP servers, or routers could be altered to run the most up-to-date code without having to be shut down at all for three years or more.

“What’s very impressive is that the work is based on very strong theory,” comments Tom Ball of Microsoft Research, “and they took it all the way to robust implementation.” Another example of a tool that Foster and Hicks have built is a program they call Locksmith. For decades, computers became faster and faster as, true to Moore’s law, hardware developers packed ever more transistors into computer processors. But in recent years, the speed of individual processors has plateaued as they have reached the physical limits for transistor density. Packing more transistors together makes processors run too hot and use too much power. As a solution, computers are being made with two processors instead of just one.

“Very soon, we will have computers with hundreds of processors and then thousands,” says Hicks. The use of parallel processors fundamentally changes how programmers have to write code, say Hicks and Foster. Software run on parallel processors must be able to integrate calculations that happen separately and make sure that information is transferred between—or among—processors in the proper order. When information isn’t processed in the right order, “data races” can cause catastrophic system failures. Indeed, the August 2003 electric blackout in the Northeast United States resulted from a data race in a power plant’s management software.

Locksmith can analyze software to ensure that it uses proper synchronization techniques for avoiding data races. This reduces the burden on programmers who are writing for parallel processors since mistakes in synchronization can be caught automatically. Because Locksmith is open-source software, other researchers and developers can freely download, use, and even extend Locksmith.

Another tool that Hicks and Foster have developed, which they call Pistachio, is a method for testing communication protocols among computers to ensure that the protocols are reliable and secure. “There’s always a gap between the code that you actually write and the idealized version of it on paper,” says Foster. “The idea is to take English language descriptions of what code is designed to do and translate that into Pistachio’s specification language. Then you can run Pistachio to check whether the specifications and the code match.” Pistachio runs very quickly and has been used to check several implementations of SSH2, a widely-used protocol for securely communicating between computers, such as between a home computer and an office network.

“Our interest started as technical, but we also want to have an impact,” says Hicks about their programs. To increase the usefulness of their tools, he and Foster are working with the Human-Computer Interaction Lab at UMIACS to see how adapting their programs’ interfaces could make their tools more accessible and effective. Microsoft’s Ball commends the way Hicks and Foster use real data to show that their techniques work. “They have quite a strong track record,” he says.

Networks and Links

Lise Getoor

Sometimes the best way to understand something is to see how it relates to others. Lise Getoor, an assistant professor of computer science and member of UMIACS, specializes in studying networks or how entities—whether people, events, research projects, or points on a map—are connected to each other. She examines how these connections can be used for machine learning and probabilistic reasoning.

“Traditionally, people have looked more at individuals instead of the links among them,” says Getoor. “We can label individuals based not only on their attributes but on what they’re connected to.” In other words, Getoor studies relationships. The work of “link mining” to extract information can be looked at in an abstract, mathematical way as well as in very practical terms. Getoor has developed algorithms for classifying all manner of data.

Exploring links in an academic context, she has analyzed how research publications are connected to each other through their authors and citations. From such bibliographic data, one can identify collaborative groups—one type of social group. Going a step further, Getoor has looked at how social relationships among academics, deduced from evidence such as how professors are grouped in committees organizing a conference, affect individuals’ social capital—and, in turn, publication rates. The same kind of analysis can be used to explore relationships between lawmakers and lobbyists, she says.

In another sort of link mining, collaborating with researchers at Carnegie Mellon University, Getoor has looked at the hyperlinks among the Web sites of professors and students. It turns out that one strong sign that a Web site belongs to a professor is that it does not include links to the Web sites of other professors! This is just one example of how one can deduce the roles of individuals by looking at the types of interactions they have. In another example, Getoor has analyzed how e-mail history reveals people’s roles in a company as well as the professional and social groups to which they belong.

In addition to link mining, one of Getoor’s main interests is entity resolution, or developing tools to deduce whether two separate mentions of a person, place, object, or event refer in fact to the same thing. For example, are references to a Joe Smith and another to a Joseph Smith talking about the same person? Entity resolution is also important in geography, and Getoor is collaborating with the U.S. government’s National Geospatial-Intelligence Agency to go through databases of locations and peg which mentions are redundant, referring to the same underlying place. For example, the agency receives updates from local governments and needs to distinguish whether new information refers to a place meriting a new entry or refers to an entity that already exists in the database, if under a slightly different name or mismatched coordinates. “It’s common for names or coordinates to be off,” notes Getoor.

Whether dealing with people or geographical locations, entity resolution requires algorithms that assess the commonalities and distinctions between two records. Clearly, if two entities’ attributes and relationships do not overlap extensively, they are more likely to be distinct rather than duplicates.

Getoor speaks about concrete problems but also about the abstract essence of those problems. “I do like the theory and algorithms as well as the compelling applications,” she says. At the level of theory, she talks about statistical relational learning and link mining driving machine learning, artificial intelligence, and reasoning under uncertainty. At the level of applications, she talks about entity resolution, group discovery, and role discovery.

“Lise has a unique way of breaking down problems and teasing out the essence of a question,” says Chris Diehl, a senior research scientist at the Johns Hopkins Applied Physics Laboratory who has collaborated with Getoor.

For her research, Getoor receives funding from the National Science Foundation as well as a consortium of intelligence agencies. “I’m also interested in how all this relates to privacy and what you can say about the privacy you can guarantee,” she says. In some cases, such as with health or financial records, the goal is to guarantee that no one can know that two records refer to the same person. In other cases, distinguishing entities can be helpful—for example, to thwart financial identity theft. In early 2007, Getoor received a Google Research Award for her work on entity resolution.

Ultimately, Getoor’s work revolves around graph identification—or taking noisy, redundant data and identifying the true underlying network of nodes and links. Getoor works to distinguish, label, and rank nodes by better assessing, categorizing, and inferring the links among them, at the same time always asking exactly how much and what kind of data is necessary to get accurate and reliable analysis.

Biology

Canna scallion leaves swiss chard, spring onion. Ceylon spinach scallion courgette water chestnut kale runner bean, spinach endive, yardlong bean. Endive; chinese artichoke courgette, kohlrabi catsear kale ulluco scallion cucumber.

Avocado fiddlehead arracacha moth bean, peanut garden rocket? Ginger radicchio chicory tarwi zucchini, bitterleaf garden rocket. Wild leek yam; burdock, summer purslane winged bean pea; beetroot samphire.

Celeriac yam mizuna greens kurrat. Polk yam; chinese cabbage, samphire tarwi, samphire lentil broadleaf arrowhead. Kai-lan broccoli rabe napa cabbage good king henry leek chinese cabbage, black-eyed pea. Sweet Potato or Kumara new zealand spinach dandelion.

Broccoli Rabe fluted pumpkin tatsoi; yarrow lizard's tail, asparagus mustard cassava cress black-eyed pea. Florence fennel green bean winter purslane.

Azuki bean, sweet potato or kumara epazote lamb's lettuce elephant garlic wild leek. Ginger, scorzonera. Garden Rocket fluted pumpkin soybean cress paracress, catsear. Pak choy garbanzo shallot ti. Hamburg parsley; parsnip yam guar brussels sprout scallion fava bean arracacha greater plantain manioc skirret canna. Swede; ulluco, plectranthus jerusalem artichoke pea sprouts turnip greens wild leek, mooli fluted pumpkin.