Computer and Information Science

Off

Mon, 12/16/2024 - 16:55

Three faculty members from three ��-Dearborn colleges — COB, CECS and CASL — earned awards to work in Egypt, Ethiopia and Indonesia.

��-Dearborn faculty members, from left, Mercedes Miranda, Kristian Stewart and Birhanu Eshete

Reckoning with AI’s trust issues

jpow — Fri, 26 Jul 2024 19:30:40 +0000

Reckoning with AI’s trust issues jpow Fri, 07/26/2024 - 15:30

It seems to be an unbreakable human habit to unleash technologies into the world and ponder the consequences later. The most recent dramatic example of this phenomenon is the artificial intelligence boom. For years, critics — even many who believe the technology can do a lot of good in the world — have been sounding the alarm that AI applications are creating sticky situations when it comes to safety and fairness. Autonomous vehicles have injured and killed pedestrians. Algorithms used in healthcare, banking and law enforcement are embedded with human racial biases. More recently, companies behind generative AI models like ChatGPT have apparently developed their models using an unfathomable amount of copyrighted material — without bothering to ask content creators.

AI’s reach into our lives is bound to get broader, and Associate Professor of Computer and Information Science Birhanu Eshete wants his students to be prepared — not only to create powerful problem-solving technologies, but to reckon with the implications for our lives. That’s the mission behind a new trustworthy AI course that Eshete created and launched in Winter 2024, which blends a project-based curriculum focused on specific technical AI applications with explorations of the technology's social, political and economic contexts.

With a two-pronged approach and a sprawling topic, it wasn’t the easiest course to design. Eshete says trustworthy AI is a label that ecompasses a diverse set of issues, from security, privacy and safety to transparency, ethics, fairness and regulation. Any one of those seven topic areas he used to organize his class could be a course on its own, he says. “But the value of an overview course is that students get introduced to a pretty comprehensive set of issues — and then they can go deeper into anything that interests them,” he says. Plus, by reinforcing that almost anything AI touches has human implications, he can drill home to a bunch of future AI professionals that AI is hardly just a technical discipline.

For example, in the ethics section of the course, the class pondered looming real-world applications of AI like autonomous trucking. Eshete says proponents of long-haul autonomous trucks often advance a safety argument — namely, that highway accidents involving trucks are often catastrophic and a result of human error. If the job could be done more precisely by machines, then we could reduce the number of injuries and fatalities. “So it might sound ethically right to make trucks autonomous, because we could save lives,” Eshete explains. “But on the flip side, you’re going to push thousands of truck drivers out of a job, and they and their families will suffer. In fact, the World Economic Forum estimated that roughly 75% of jobs have the potential to be replaced by robots and artificial intelligence. Imagine the disruptions to society. So workforce replacement is not just a technology issue. It’s a human issue.”

You can also see the complexity and variety of the trustworthy AI space in the range of topics students chose for their final projects. One team looked at algorithms used to detect skin cancer to analyze the technology's level of racial and gender bias and explore ways to enhance fairness while protecting privacy. Another group explored techniques for reducing the error rate of stop sign detection algorithms used in autonomous vehicles. A third team investigated techniques used to acquaint a model with tricky inputs designed to fool it. And the final group of students explored how to “poison” databases of customer reviews, a technique attackers use to influence consumer sentiment about products.

Doctoral student Firas Ben Hmida, who’s also a research assistant in Eshete’s lab, says the trustworthy AI topic has been a real eye opener for him. “When I started at my engineering college back home in Tunisia, I started off in cybersecurity, and my interest, probably like everybody else, was advanced attacks and how to defend against attacks,” Hmida says. “But then, as I started working with Professor Eshete, you see that there is this whole other side to cybersecurity and AI. Outside the course, even several of the projects we’re working on in the lab deal with trustworthy AI, so I’m even thinking my PhD is now going to be about trustworthy AI.”

Eshete with his doctoral students Firas Ben Hmida (right) and Philemon Hailemariam. Photo by Erin Kirkland/Michigan Photography

Eshete heard that kind of feedback pretty consistently from students. “I got a lot of comments like, ‘It allowed me to look at AI beyond neural networks,’” Eshete says. “That’s great to hear because you don’t know how students are going to react when you spend half your lecture in an AI course talking about Socrates, Plato and Aristotle and ethical philosophy.” In fact, Eshete says the feedback was the most positive of any course he teaches. Specifically, he says students appreciated that the overview format allowed them to explore many topics in a single course. They also liked the project-based format, which eliminated the need for exams and quizzes.

Eshete plans to teach the class on campus every winter semester, with the exception of Winter 2025 when he’ll be on sabbatical. He’ll actually still be teaching the trustworthy AI course though — just at Addis Ababa University, his alma mater in Ethiopia, where he'll be spending part of his leave. “I’m actually really looking forward to teaching the course in two different countries, because any of these AI trustworthiness issues could mean many different things depending on the context,” he says. “For instance, if you’re thinking about fairness and race here in the U.S., we’re accustomed to thinking about things in terms of Black and white. But in Ethiopia, everybody is Black. So the context of race is slightly different, and we think more about different ethnic groups. So it will be interesting to see how students respond to thinking about these issues for AI applications in high-stakes contexts.”

###

Story by Lou Blouin

Careers or Internships

Experiential Learning

Off

Mon, 07/22/2024 - 19:30

In his new trustworthy AI course, Associate Professor Birhanu Eshete is helping students go beyond the technical know-how and understand the social, political and economic contexts of the artificial intelligence boom.

Photo by Erin Kirkland/Michigan Photography

��-Dearborn is infusing sustainability into its engineering programs

lblouin — Wed, 03 Jul 2024 12:57:45 +0000

��-Dearborn is infusing sustainability into its engineering programs lblouin Wed, 07/03/2024 - 08:57

The days when engineering was thought of as a purely technical discipline are disappearing fast. For sure, engineers still need strong foundational knowledge in math and science and discipline-specific technical training. But there’s a growing movement within engineering education to ensure we’re training engineers who can also think deeply about the environmental and societal impacts of their creations as a fundamental part of the engineering process.

At ��-Dearborn, this emphasis on creating holistic engineers is already taking a number of forms, from sustainability-focused classroom projects and research, to a new course in trustworthy AI, to undergraduate and graduate programs in human-centered design, a discipline that blends traditional engineering with skills borrowed from art, anthropology and sociology. Now, sustainability is poised to take an even broader role within the College of Engineering and Computer Science, say Associate Professors Alireza Mohammadi and Samir Rawashdeh. The robotics experts and frequent collaborators recently landed a $200,000 grant from the Lemelson Foundation to support CECS faculty who want to incorporate sustainability-focused projects into their courses. “In talking with the people at Lemelson, we realized if you want this to become part of the culture and not just an occasional or fringe thing, you have to reach a critical mass,” Rawashdeh says. “Sustainability has to be something students are hearing about in multiple places, across courses, across programs for them to internalize that this is part of what engineering is all about.”

Specifically, the Lemelson funding will enable Mohammadi and Rawashdeh to award minigrants to faculty who want to add course modules or projects that incorporate the Engineering for One Planet Framework, a set of fundamental environmental sustainability-focused learning outcomes developed by Lemelson and VentureWell, with input from stakeholders in academia, industry and nonprofit organizations. For example, in a smaller-scale antecedent to this larger Lemelson-funded initiative, Rawashdeh and Mohammadi used a minigrant to add sustainability-focused modules to two of their robotics courses. “It was an interesting challenge because the overlap between robotics and environmental issues isn’t obvious,” Rawashdeh says. But as they read up on it, they got all kinds of ideas — from robotics applications for environmental industries, like recycling, to end-of-life e-waste management strategies for robotics components, including their valuable lithium-based batteries. In Mohammadi’s course, the students optimized control algorithms for their lab’s fleet of mobile robots so that the machines would consume less energy while doing the same amount of work.

While the primary goal of the initiative is to develop a more environmentally focused mindset among engineering students, Mohammadi says emphasizing sustainability in CECS programs could also make engineering disciplines more attractive to a broader range of students. “The younger generation cares more about environmental and social issues, and engineering is absolutely a profession where you can make a large impact on society,” Mohammadi says. “But we don’t often describe it that way. So showing them how they can shape the world in a way they care about might make students who’ve never considered engineering before think about it as an option.” To match students and employers, they’ll also be hosting several industry events, where employers can talk with students about their sustainability-focused opportunities and students can showcase their work.

Interested faculty can apply now for minigrant funding by contacting Mohammadi or Rawashdeh. They’re hoping to fund projects in 10-16 courses over three years.

###

Story by Lou Blouin

Faculty and Staff

Nature or Environment

Electrical and Computer Engineering

Industrial and Manufacturing Systems Engineering

Mechanical Engineering

Off

Wed, 07/03/2024 - 12:55

Part of being a great engineer these days is understanding the environmental and societal impacts of your work. But how do we train students to be holistic engineers?

Graphic by Violet Dashi. Images by Marianna and KUA g Gear via Adobe Stock

COB, CECS online programs among top in the country

lblouin — Wed, 07 Feb 2024 17:49:37 +0000

COB, CECS online programs among top in the country lblouin Wed, 02/07/2024 - 12:49

Earlier this academic year, U.S. News & World Report released their annual guide to American colleges and universities and ranked ��-Dearborn among the best in the Midwest — listed as the No. 2 regional public university.

In rankings released Wednesday, U.S. News again included ��-Dearborn in their rankings of more than 1,600 online programs.

The College of Business’ online MBA and online master’s in business programs (excluding MBA) and the College of Engineering & Computer Science’s online master’s in Information Systems and Technology were named among the best. Curious about the methodology? U.S. News explains it.

CECS’ online graduate degree in Information Systems and Technology is the second-ranked program in Michigan and No. 44 overall.

COB also ranked as second in the state and No. 68 overall on the list of Best Online Master’s in Business Programs. This isn't the first time this academic year that COB programs have been recognized. In fall 2023, Poets&Quants, a leading online publication for business education news, ranked the online MBA as No. 28 in the United States — second in Michigan, only behind ��-Ann Arbor’s Ross School of Business — and called ��-Dearborn’s program a Best Online MBA Program of 2024. Poets&Quants also named COB a Best Undergraduate Business School for 2024.

COB’s online MBA program — ranked in the No. 3 spot in Michigan on U.S. News and World Report's Best Online MBA Programs list — is also recognized on the top program for veterans list, which showcases universities that make pursuing a college education more affordable for veterans and active-duty service members.

In the fall, U.S. News named ��-Dearborn as the No. 1 in the state on the “Best College for Veterans” list among regional public universities. Criteria for the designation include being GI Bill-certified and participating in the Yellow Ribbon Program. In addition, ��-Dearborn has veterans-based programming and provides structure for incoming Dearborn Wolverines who have served.

###

If you are a service member or veteran interested in learning about the campus or degree programs, contact Veterans Affairs Coordinator Tom Pitock. Article by Sarah Tuxbury.

Academic Excellence

Online Learning

College of Business

COB Graduate Programs

Off

Wed, 02/07/2024 - 17:01

U.S. News & World Report’s rankings include ��-Dearborn’s outstanding online programs and ranks campus’ MBA as a top graduate program for veterans.

A graduate student presents in the Bloomberg Lab, one of the College of Business' flagship facilities. Photo by Julianne Lindsey

This competition is hoping to address the shortage of EV battery engineers

lblouin — Mon, 29 Jan 2024 17:43:54 +0000

This competition is hoping to address the shortage of EV battery engineers lblouin Mon, 01/29/2024 - 12:43

You need only look at commercials to know that automakers are betting big on electric vehicles. Indeed, nearly all major vehicle manufacturers have announced significant shares of their fleets will go electric by the end of the decade, with more than a handful announcing plans to completely phase-out combustion technology in that timeframe. Not surprisingly, the rapid shift toward EVs isn’t arriving without a few growing pains. One of the biggest current challenges for automakers is simply finding enough engineers who are well-versed in the new technologies, especially the high-capacity lithium-ion batteries that are fueling the EV revolution. A recent report from Bloomberg noted that the workforce shortages are so acute that a senior-level battery engineer can command a salary on par with a company CFO. In Japan, a government-industry partnership has turned to educating high school students for battery-related careers, in an effort to train the 30,000 workers they’ll need to keep the industry thriving.

The U.S. Department of Energy has also taken note, launching a variety of initiatives to help sustain the EV industry and reach the U.S. government’s goal of net-zero carbon emissions by 2050. Among them is the Battery Workforce Challenge, a national competition co-sponsored by the Department of Energy and Stellantis and managed by Argonne National Laboratory, which challenges collegiate student teams to design, build, test and integrate an advanced EV battery system into a future Stellantis electric vehicle. It’s an incredible opportunity for students, says Associate Professor of Electrical and Computer Engineering Xuan (Joe) Zhou, who’s advising the ��-Dearborn team, one of 12 teams, and the only team in Michigan, selected for the three-year competition. “Stellantis is supplying a real battery cell, the charger, even a vehicle to us, and then the students will design the whole pack, including the battery management system, and put it in a real vehicle,” Zhou says. “I’ve been working with batteries for many years, and even I haven’t had a chance to work with a whole pack! So this is really going to be special for the students.”

Because of the focus on workforce development, the competition is also structured to closely mimic the industry environment, which includes phases for design, simulation, controls development, testing, and vehicle integration and demonstration. Teams will periodically report in with key deliverables, which gives them an opportunity to get feedback on their work, just like they would in a real workplace. Zhou says each team is assigned two Stellantis engineers, who they can consult with during the competition. Two students from each team will also get to complete internships with Stellantis.

Zhou is also excited that the ��-Dearborn team is enlisting the talents of students from neighboring Henry Ford College. In Fall 2023, HFC became the first college in the country to offer a Battery Technician Certification curriculum, which is part of its growing lineup of vocational programs focusing on electric vehicles. “I think the DOE is really excited to see us working with HFC because, of course, we need engineers to design the vehicles, but we also need high-skill technicians who can fix the vehicles,” Zhou says. “Also, we have a very close relationship with HFC, where students with a certain GPA are guaranteed admission to ��-Dearborn. So for HFC students who want to transfer to a four-year engineering program, it’s an excellent opportunity to come to our campus and get hands-on with the latest technology.”

��-Dearborn doctoral student Duncan Tyree, whose current research is focused on battery manufacturing, says he joined the Battery Workforce Challenge team to deepen his expertise in one of his favorite subjects. But he’s also hoping the broad scope of the competition will stretch some of his non-engineering skills. As the ��-Dearborn team’s project manager, he’s looking forward to mentoring the younger engineering students, as well as to opportunities for sharpening his communication and meeting facilitation skills. Moreover, because his academic experience has focused mostly on simulation, he’s excited to work on a project that will ultimately have a real-world proving ground. “I think year two is going to be the most challenging — that’s when we actually start putting the battery pack together,” Tyree says. “In some ways, for someone with my background, doing things like putting a battery in the vehicle could be the least exciting thing. But like with any project, there are times when a manager has to step in and help out with whatever needs to get done, and I’m actually really looking forward to that part of it. I’ve had some leadership positions, like with student organizations, and this feels a lot different. This is a national competition, and so there’s a lot more at stake. People are counting on you, and we all want to do as well as we can.”

Zhou and Tyree both say recruiting for the team is going well. The group now includes more than 20 ��-Dearborn students from a variety of programs and another 10 students from Henry Ford College — though Zhou says there’s still plenty of room for more. “Basically, we can’t have enough help,” Zhou says. “It’s a very complex project, and we are up against some top engineering schools like the University of Alabama and Ohio State, who have years of experience doing competitions like this. But I think we have the makings of a strong team. They’re hard-working students and I’m very excited for them to have this opportunity.”

###

Are you interested in joining ��-Dearborn’s Battery Workforce Challenge team? If so, please reach out to project manager Duncan Tyree or faculty advisor Associate Professor Xuan (Joe) Zhou. You can also learn more about the Battery Workforce Challenge on the Advanced Vehicle Technology Competitions website. Other faculty members supporting the Battery Workforce Challenge team include: Professor Wencong Su, Lecturer Azeem Hafeez, Assistant Professor Jaerock Kwon, Associate Professor Maggie Wang, Assistant Professor Junho Hong, Associate Professor Samir Rawashdeh, Associate Professor Youngki Kim, Assistant Professor Lei Chen, Assistant Professor Zhen Hu and Assistant Professor Zheng Song. Story by Lou Blouin

Student Success

Electrical and Computer Engineering

Industrial and Manufacturing Systems Engineering

Mechanical Engineering

Off

Mon, 01/29/2024 - 16:56

The national collegiate Battery Workforce Challenge competition is giving ��-Dearborn and Henry Ford College students a rare opportunity to train on real EV batteries and real vehicles, while preparing them for in-demand careers.

From left, Julian Summers, Duncan Tyree, Steve Su and Nicholas Wheelock are among the more than 20 ��-Dearborn students who've joined the Battery Workforce Challenge team. Photo by Emily Barrett-Adkins

Class of Fall 2023: CECS graduate Olivia Pellegrini

lblouin — Mon, 11 Dec 2023 15:53:35 +0000

Class of Fall 2023: CECS graduate Olivia Pellegrini lblouin Mon, 12/11/2023 - 10:53

High school robotics clubs or computer science classes provide many students with a gateway to a college experience in a STEM field. Unfortunately, College of Engineering and Computer Science senior Olivia Pellegrini’s small Downriver high school didn’t have anything like that. What she did have was a big brother who looked out for his younger sister. When she was in high school, he was studying computer engineering at ��-Dearborn, and recognizing his sibling’s love of math, he started teaching her all the stuff he was learning in his introductory computer science courses. “I guess we sort of had our own little computer science club at my house, just the two of us,” Pellegrini says. The brotherly mentoring had an impact: When Pellegrini graduated, she followed in his footsteps and enrolled at ��-Dearborn, where she zeroed in on a major in computer science.

After working her way through her introductory classes, Pellegrini discovered a passion for cybersecurity. She loved that the courses were often built around real-world projects. In one of Professor Di Ma’s classes, for example, they dug into the security vulnerabilities of actual Android apps. In a course with Associate Professor Anys Bacha, the students got to recreate the infamous Samy worm attack on MySpace, one of the seminal attacks on a social media platform. “It was pretty cool to play the role of attacker and actually see how the attacks work, because that’s how you learn to defend against them,” Pellegrini says. “It’s one thing to learn the concept from a book, but when you can actually implement it, it sticks, and it’s there now, like a tool in your toolbox that you can pull out whenever you need it.”

The practical, hands-on learning continued with Assistant Professor Zheng Song, who tapped Pellegrini and her senior design teammates for a research project focusing on edge computing applications for unmanned aerial vehicles or UAVs. Their project, dubbed Sky Socket, involved working on a drone-vehicle communication system. The system allowed drones to conduct advanced reconnaissance and relay information back to ground vehicles, giving emergency or military personnel the ability to have eyes on the ground far beyond their actual lines of sight. The work was so successful that Song, Pellegrini and her teammates presented the results in a paper at the IEEE: MOST conference, an international gathering of experts focusing on advanced mobility technologies. After senior design, Pellegrini continued to work in Song’s lab, this time on a project focusing on API key leakage, an emerging cybersecurity threat in the app-based software space. Their analysis of GitHub software repositories revealed that some 40,000 API keys, a vital security feature, are revealed in plain text. They’ve submitted that work for publication as well.

Pellegrini (far right) with Song (far left) and the rest of the Sky Socket senior design team, from left to right, Demetrius Johnson, Ryan Sauer and Jonathan Schall. (Photo courtesy Olivia Pellegrini)

Two internships gave Pellegrini yet another avenue for hands-on learning. Most recently, she worked at Marathon Petroleum in Detroit, where she created visual dashboards that helped the company’s engineers monitor critical systems. And during COVID, Pellegrini scored a spot on the team that manages FordPass, the smartphone app for the automaker’s connected vehicle features like remote start. That opportunity gave her some valuable remote work experience, though it did come with one nice in-person perk: During the final week, all the interns got to test drive the electric Mustang Mach-E and the Ford F-150 Lightning.

On top of all this, Pellegrini even found time to play two years of volleyball, an experience she never imagined she’d have at the college level. In fact, as she looks back on her college career, she’s surprised herself in a lot of ways. “I definitely think that through doing all these extracurriculars, like research and internships, it’s really boosted my confidence,” she says. “I was actually kind of shy before. And especially through playing volleyball and hanging out with those girls, it made me a more social person, so then I wasn’t afraid to do something like contact a professor about a research opportunity or talk to an employer at a career fair. Now, because of the research I did with Professor Song, I see myself working in the automotive field, especially something involving computer vision or artificial intelligence. That’s not something I would have even dreamed about heading into college.”

With a resume like hers, it’s a dream that’s certainly within reach.

###

Story by Lou Blouin

Commencement

Student Success

Undergraduate Research

Off

Mon, 12/11/2023 - 15:48

During her time at ��-Dearborn, the cybersecurity major has scored big internships, published research and even found time to play on the volleyball team. Now she’s eyeing a career in the artificial intelligence field.

Graphic by Violet Dashi

Is AI really a threat to human civilization?

lblouin — Mon, 26 Jun 2023 16:45:14 +0000

Is AI really a threat to human civilization? lblouin Mon, 06/26/2023 - 12:45

Even if you haven't been following the current conversations around artificial intelligence, it’s hard not to do a double take at the recent headlines warning that AI may soon represent a serious threat to human civilization. The stories revolve around recent statements made by several industry leaders, including one of AI’s “godfathers,” that the technology is now evolving so rapidly, it could represent an “extinction-level” threat sooner than expected — or at least trigger societal-scale disruptions on par with the recent global pandemic. Frankly, when you’re not an AI expert yourself, it can be hard to know what to make of such claims. For sure, today’s incarnations of AI can do impressive things, including many things humans could never do. And as we’ve detailed in the past, current AI-powered technologies have a problematic track record, whether it’s amplifying disinformation, perpetuating human racial biases or supercharging criminal scams. Maybe it’s not so hard to believe that the machines really are about to get the better of us.

Here at ��-Dearborn, we have a lot of faculty who work with artificial intelligence, so we thought it’d be interesting to put the question of whether AI really is on the verge of becoming a civilization-ender to three of the university’s leaders in this area. Professor Hafiz Malik, Associate Professor Samir Rawashdeh and Assistant Professor Birhanu Eshete were all in agreement that the assertion that artificial intelligence could represent an extinction-level threat any time soon is overblown. Rawashdeh quipped that folks seem to be forgetting that “we can always unplug them if they start misbehaving.” Eshete generally rates today’s version of artificial general intelligence as “cat-level.” Malik said that’s not giving cats enough credit. “I think when you see some of the impressive things AI can do, it can be easy to get the impression that the technology is further along than it is,” Malik says. AI may be able to beat the best human chess player or diagnose illnesses doctors can’t, but he notes those are “very task-specific things with strict constraints.” “General intelligence, the kind of intelligence that humans possess, where we can adapt to new circumstances, that’s not a kind of AI I would expect to see in my lifetime,” he says. “And as far as the pace of advancement, I would say it has been fairly steady, not accelerating rapidly.”

This points to an important distinction between types of artificial intelligence that’s often overlooked in the current discussions around AI. Today’s AI, which is mostly driven by machine learning, is task specific. It’s the technology that allows algorithms, when given enough exposure to, say, photos of cars or X-rays of a particular type of cancer, to learn the essential characteristics of those things. But artificial general intelligence, or AGI, is a completely different creature. It would mean that machines could, as humans do, adapt to an almost infinite array of new tasks without being specifically trained or programmed to do those tasks. Notably, while task-specific AI is becoming ubiquitous, AGI doesn’t exist yet. Some doubt that it’s even possible. If it is achievable, there are many who think that it wouldn’t look anything like today’s AI.

Still, Rawashdeh, Eshete and Malik all say it wouldn’t take something as advanced as AGI to cause big problems in the human world. Rawashdeh and Eshete both voiced concerns over the fact that the highest levels of artificial intelligence are basically controlled by a handful of very large, powerful companies, which are developing the technology for commercial purposes, not to benefit human society. “I think the real risk is we could very quickly become dependent on the technology in a huge range of sectors, and then we end up with systems that perpetuate inequality,” Rawashdeh says. “And at that point, you could imagine people saying, ‘Well, you can’t just turn it off because it would crash the economy.’” Like the justification used to bailout misbehaving banks during financial crises, AI could be judged too big to fail.

Disinformation is the other obvious area where we’re already seeing AI’s disruptive power. Disinformation, of course, is likely as old as human civilization itself. But Malik, who’s an expert in deepfakes, says AI has supercharged its impacts. “The polarization which we’re seeing around the world, not just in the U.S., has a lot to do with social media platforms, which are driven by algorithms, creating echo chambers where people end up with very distorted views of reality,” Malik says. Deepfakes, which he says are “getting better and better every day,” have only made people more vulnerable. In fact, scammers are now putting deepfake technology to use in even more clever ways. Malik says criminals can now use AI to synthesize voices in real time, complete with an array of accents, powering convincing phone scams designed to scare people into draining their bank accounts. Whether it’s social media disinformation or a criminal scam, Malik says the result is a general erosion of trust in information and democratic institutions. And if we’re looking for things that could legitimately contribute to an unraveling of human society, this loss of trust seems like a good place to start.

Concerns over these problematic sides of AI technology have also sparked conversations about how to protect ourselves, and Eshete and Malik say the European Union has been a leader when it comes to regulation. Just this month, the European Parliament passed a draft version of the EU AI Act, which, among other things, would seriously limit the use of facial recognition software and require creators of generative AI systems like ChatGPT to be more transparent about the data used to train their programs. Here in the U.S., Eshete notes the White House has also released the AI Bill of Rights to “help guide the design, use and deployment of automated systems to protect the American Public.” Eshete says penning good regulations is complicated by the fact that there still is no consensus among AI experts on whether AGI systems have anything resembling human capabilities, or even which risks we should be most worried about. He notes it’s really easy to get distracted by the frightening, future hypothetical threats of AI, like creating a lethal bio-weapon. “But there are all sorts of ways AI is already impacting people’s lives. So perhaps we should focus first on what is happening right now. And then once we’ve done that, we’ll have time to look at what’s coming.”

Eshete, Rawashdeh and Malik all say how much AI ultimately ends up reshaping our world, and whether its impacts will be beneficial or harmful, is largely up to us. Could we end up in a place where AI really does become a civilization-ender? Possibly. But if we do, we likely won’t have the machines to blame.

###

Story by Lou Blouin

Electrical and Computer Engineering

Off

Mon, 06/26/2023 - 16:44

Three ��-Dearborn experts say recent warnings that artificial intelligence may soon pose an ‘extinction-level’ threat are overblown. But we’re right to be concerned about other ways AI is reshaping our world.

In the HBO series "Westworld," AI-powered androids from a Western theme park turn the tables on humans and turn us into the puppets. Credit: John Johnson/HBO/Warner Bros.

Two ��-Dearborn students awarded prestigious Rackham Predoctoral Fellowship

lblouin — Mon, 15 May 2023 16:03:34 +0000

Two ��-Dearborn students awarded prestigious Rackham Predoctoral Fellowship lblouin Mon, 05/15/2023 - 12:03

Back in 2019, the four doctoral programs in ��-Dearborn’s College of Engineering and Computer Science kicked off a formal partnership with the Rackham Graduate School at ��-Ann Arbor. The Rackham affiliation did more than just add some additional cachet. Brahim Medjahed, CECS Assistant Dean and our campus liaison with the Rackham Graduate School, says it’s helped us strengthen those Ph.D. programs in fundamental ways, including adding evidence-based strategies for effective mentoring and a curated lineup of professional development opportunities. Perhaps most importantly, it’s now a standard requirement that all doctoral students be fully funded with a graduate assistantship — a feature that’s designed to help students finish their degrees as quickly as possible, while their research and dissertation work is still relevant.

Graduate assistantships, which require doctoral students to have campus employment duties such as teaching courses, are a key part of making doctoral programs affordable. But teaching classes while conducting research, taking doctoral courses and working on a dissertation can obviously be a lot to handle. One thing that can be a big help: the Rackham Predoctoral Fellowship, a competitive program that provides a $36,084 stipend, plus doctoral candidacy tuition and fees, for especially promising doctoral students in their final year. This means they can forgo their regular instructional responsibilities and focus solely on their research and dissertation. “That last year is so crucial,” Medjahed says. “It’s when students start publishing research results, finish their dissertations and start looking for jobs. So the fellowship puts students in a much stronger position to launch their careers. And it’s also obviously a great thing for a resume because it’s a highly competitive and prestigious fellowship.”

The Rackham Predoctoral Fellowship is traditionally awarded to around 100 doctoral students on the Ann Arbor campus. But this year, for the first time, two students from CECS’s doctoral programs were named to the list of fellows. Look at their work and it’s easy to see why Computer and Information Science student Nada Lachtar and Industrial and Manufacturing Systems Engineering student Dania Ammar were given the nod. Both are working on high profile problems: Lachtar is developing broadly applicable, novel solutions to ransomware attacks, while Ammar is designing human-machine interfaces that will help pedestrians interact safely with autonomous vehicles. As doctoral students, both have published in high quality journals and presented at top conferences. And even with their busy schedules, both have found time to serve on university committees and mentor ��-Dearborn students coming up behind them.

Lachtar says she first developed an interest in ransomware when she was an undergraduate student in 2016, the year ransomware blossomed from a niche attack strategy to a major cybersecurity threat. At the time, Lachtar was looking into graduate programs and ended up connecting with ��-Dearborn Assistant Professor Anys Bacha, who was also working in this area. Over the past five years, they’ve tackled a broad range of ransomware related issues. One of Lachtar’s earliest projects was a machine learning threat detection system that was 99% effective at recognizing malware. Later, she developed operating system-level defenses that were capable of shutting down attacks and limiting damage. She’s also worked on strategies for detecting adversarial machine learning, a type of attack that actually uses machine learning to sidestep machine learning-based defenses.

Ammar’s research is exploring one of the more fascinating emerging challenges related to autonomous vehicles: How will pedestrians safely interact with cars when humans aren’t doing the driving? Ammar says we typically rely on visual signals, like eye contact, to negotiate who has the right of way, and absent a human driver, cars will presumably have to have some way of declaring their intentions. For example, through her research, Ammar has discovered that pedestrians tend to favor technologies that closely mimic their interactions with human drivers — say, a visual or auditory message, delivered from the car, that says “I see you! It’s safe to cross.” Ammar’s dissertation work is focused on resolving many such design complexities, including the incredibly murky territory of “trust calibration.” Essentially, she says, keeping traffic moving efficiently and keeping humans safe rely on pedestrians being both trusting of vehicles and sufficiently wary so they can adapt to surprises. Designing systems that encourage this “sweet spot” behavior will be just as key to making autonomous vehicles a reality as the core technology now being developed by automakers.

As someone who’s been key to building CECS’s doctoral education culture, including the affiliation with Rackham, Medjahed is proud to see students from the ��-Dearborn campus earning this recognition. “Both Dania and Nada are extremely deserving,” he says. “You look at what they’ve accomplished — the research maturity, the level of productivity and their engagement with the university — they’re everything you’d expect from the top Ph.D. candidates in top Ph.D. programs around the country.” Mike Solomon, dean of the Rackham Graduate School, also sees it as an important milestone for the Dearborn campus. “The Rackham Predoctoral Fellowship is an important part of our school’s rich legacy,” Solomon says. “That legacy has been enhanced by the addition of Ph.D. programs at ��-Dearborn, which are now fully integrated with the graduate school. I am therefore very pleased to welcome this inaugural cohort of Dearborn Rackham Predoctoral Fellows, and I thank the many staff and faculty on both the Dearborn and Ann Arbor campuses who have worked to make this opportunity possible.”

###

Story by Lou Blouin

Industrial and Manufacturing Systems Engineering

Off

Mon, 05/15/2023 - 16:02

Nada Lachtar and Dania Ammar are the first ��-Dearborn doctoral students to be recognized with the competitive fellowship from one of the country’s premier graduate schools.

Nada Lachtar (left) and Dania Ammar were recently awarded the Rackham Predoctoral Fellowship.

Cybersecurity researcher Birhanu Eshete scores prestigious NSF CAREER award

lblouin — Wed, 05 Apr 2023 13:08:30 +0000

Cybersecurity researcher Birhanu Eshete scores prestigious NSF CAREER award lblouin Wed, 04/05/2023 - 09:08

The young talent in ��-Dearborn’s College of Engineering and Computer Science continues to impress big national funders. Along with fellow Assistant Professors DeLean Tolbert Smith and Fred Feng, Birhanu Eshete can now count himself among the CECS faculty who’ve recently landed the National Science Foundation’s CAREER Award, a prestigious nationwide grant given annually to about 500 early-career faculty who are emerging leaders in their fields.

You don’t get the nod for a CAREER Award with boilerplate ideas, and Eshete will be going after some big fish with his five-year $619,838 grant. Specifically, he’ll be exploring a new approach to machine learning’s “robustness” problem, which continues to be one of the biggest challenges in artificial intelligence, primarily because it’s so intimately tied up with the safety and security aspects of AI systems. So what exactly is robustness? Within AI disciplines, robustness can actually refer to a few different things. If we’re talking about an autonomous vehicle, for example, we’d say its navigation system is robust if it can respond correctly to slightly novel situations it’s never seen before, like stopping at a stop sign even if there’s a minor glare or a little snow on it.

Alongside safety, the other area where robustness is a big deal is cybersecurity, which is Eshete’s specialty. In this space, robustness primarily refers to whether a machine learning system can protect itself from a range of attacks. For example, if you’re a medical researcher building an AI-powered tool that can diagnose cancer by looking at a patient’s medical scans, your system was almost surely trained on real people’s medical information. Because of this, protecting people’s privacy is a big concern, especially as other doctors and hospitals start using your tool in the real world. Eshete says if your model isn’t sufficiently protected, it could be susceptible to an “inference attack” in which the model can be probed to identify a particular person whose data was used to train your model. In another kind of attack, Eshete says a malicious actor could provide scans that look legitimate, but contain “noise” specifically designed to manipulate the algorithm’s “decision boundary.” By intentionally “poisoning” the system with bad data, the model’s high-stakes medical diagnoses could become inaccurate.

Eshete says AI practitioners typically use two kinds of strategies to make models robust against attacks. You can make the model harder to attack; or, you can try to “clean” the inputs of malicious “noise” before they get into the system, sort of like an antivirus pre-scan of email attachments. But Eshete thinks both of these approaches are ultimately limited, as attackers will invariably find novel ways to thwart new defenses. Instead, he’s proposing a more fundamental solution that borrows insights from key parts of his work on nation-state cyberattacks. Such highly sophisticated attacks evolve over longer periods of time, beginning with an infiltration point that lets a hacker gradually probe more deeply into other parts of the network, where real damage can be done. Eshete developed techniques for describing these pathways of attack, which is known in the cybersecurity world as attack provenance. “It basically gives you a narrative of how the attack unfolded, and within that you’ll find all kinds of juicy information that can help with attack detection and forensic analysis,” he says. For example, such information helped Eshete develop anomaly detection systems, which thwart attacks by using machine learning to develop models of what normal network activity looks like, so unusual activities stick out as potential threats.

For his CAREER award project, he’ll be using this idea of provenance capture and analysis as the basis for a robust cybersecurity defense strategy for AI models. Here’s how it would work: First, as a machine learning model is trained on its initial data, Eshete would capture a narrative of the model’s evolution as “training provenance,” which helps establish the typical trajectory of what happens during training. Then, as the model goes live and users are feeding it new real-world inputs, he would similarly track what happens as an input moves all the way through the model to a final predictive output, thus capturing the equivalent of a “thought process” that led to a decision. This is called “prediction provenance.”

Importantly, each kind of provenance builds a signature for how the model normally functions, and Eshete says some critical insights can be gathered when you retrain the model with new inputs, which is a typical periodic maintenance procedure for machine learning systems. “When you retrain a model for which training provenance has been established, but with newly acquired and hence potentially poisonous data, if there is enough deviation from the stabilized training provenance, then that is a reason to suspect a data poisoning attack,” Eshete says. Similarly, if a particular prediction provenance sufficiently resembles a known malicious prediction provenance signature, Eshete knows he may be dealing with an input aimed at misleading the model. If the prediction provenance is more in line with the clean provenance signature, he’s more likely dealing with a benign input.

Interestingly, Eshete’s core idea could have implications far beyond the cybersecurity realm. As we wrote about in a story earlier this year, AI’s so-called “black box” problem continues to be a major hurdle to building safe, trustworthy AI systems. The problem stems from the fact that most machine learning systems are completely opaque about how they come to their conclusions — we get the decision but not the “thought process.” Eshete’s methods of creating prediction provenance could therefore provide new insights into how models are making their decisions — giving us the potential to build more transparent, repairable, trustworthy and less-biased AI systems.

Eshete will be working on the project through 2028, which will also include an educational component for Taylor High School students in the final two years. Right now, though, he’s taking a moment to enjoy the big vote of confidence that comes with being a CAREER awardee. “It does feel like a validation of your ideas and your career trajectory,” Eshete says. “And I’m grateful to the Office of Research and my colleagues and the campus in general, who’ve been very supportive and optimistic about research on our campus. It’s a very big undertaking to apply for a grant like this — at least it was for me. You can’t operate in a vacuum, and this kind of support is invaluable for moving research forward.”

###

Want to read more about Eshete’s work? Check out our stories “A dispatch from the cybersecurity arms race” and “Should we view cyberattacks as acts of war?” which spotlight Eshete’s research on nation-state attacks. Story by Lou Blouin.

Faculty Research

Research

Off

Wed, 04/05/2023 - 13:08

Eshete’s project could have wide-ranging implications for one of artificial intelligence’s most stubborn problems.

Assistant Professor of Computer and Information Science Birhanu Eshete. Photo by Lou Blouin

This video game helps kids build their defenses to online predators

rudmehta — Wed, 30 Nov 2022 18:27:28 +0000

This video game helps kids build their defenses to online predators rudmehta Wed, 11/30/2022 - 13:27

It was beginning to develop a reputation as the unfinishable project. In fact, ��-Ann Arbor Emerging Technologies Librarian Patricia Anderson spent nearly a decade trying to fulfill her vision of a video game that could help kids better detect online predators.

A few years ago, the ambitious goal even stumped one of Computer and Information Science Professor Bruce Maxim’s senior design teams, whose attempt fizzled when they got stuck on the game art. Understandably, when Maxim said he’d found another group of seniors willing to take a run at the project this year, Anderson said she kept her hopes in check. Best case scenario, she thought she’d end up with a good enough start to launch a Kickstarter campaign.

But after eight months of steady work, the team, anchored by Sean Croskey, Luke Pacheco, Aristotelis Papaioannou and Dominic Retli, has blown through those expectations.

Today, they’ve got a fully functioning, retro-inspired, Nintendo-esque, video game (a style that just so happens to be having a bit of a revival moment right now). Even more impressive, the middle schoolers they tested it with dug in just like it was any other video game.

“When I heard the kids freaking out when they got to the bear chase, it was like, ‘Yes—we did it,’” Papaioannou said, smiling.

The roots of the project are personal for Anderson. She’s a longtime advocate for people with disabilities, who, she explained, often face greater risks online.

“Any of us can be targets,” said Anderson, who works in the Taubman Health Sciences Library. “But it’s not at all uncommon for people with a range of physical or intellectual disabilities—or even those who are homebound—to rely on the online world for a lot of their social interaction. That can be really empowering, but also risky. And we wanted to help people who are navigating in that world understand how to protect themselves, and not share too much personal information with others they don’t know.”

A video game, Anderson thought, could be a solid training ground to build those skills.

To develop the game concept, she enlisted the help of her son, who has autism, and together they worked out a story anchored by a classic video game quest. During the course of your adventures, you’d encounter a variety of characters—some good, some who just pretend to be good.

In the game Ab Errantry, you play the role of a knight helping find a princess or prince (your choice) who has gone missing. Along the way, you meet characters who can help your mission by providing vital information; and others, who appear helpful, but are really out to harm you. To figure out which is which, you engage in dialogues with the characters and make decisions. The different scenarios are meant to be an analogue for the kinds of situations that could get you into trouble online.

“For example, one character is a lady who tells you she’s lost her Siamese cat,” Pacheco said. “She’ll ask you to go through this door to help her look, but if you ask her more questions, you realize her story doesn’t quite add up.” If, on the other hand, you just trust her and walk through the door, there’s a goblin waiting on the other side.

The team carefully scripted these parts of the game, working with Adam Grandt, a information security consultant who analyzed linguistic patterns from actual chat transcripts involving online predators. But they also wanted the dialogues to feel natural.

“We had to make sure it didn’t sound too preachy,” Croskey said. “For the game to be effective, the kids have to interact with the characters as they would in real life. If they got a sense that we wanted them to answer in a certain way, they might do things just because they knew it could help them beat the game.”

One of the clever parts of Ab Errantry is players don’t even know it’s a teaching tool until the very end. That’s when a post-game wrap-up helps you analyze the decisions you made, giving you constructive criticism about what you could have done better. Until then, it pretty much just feels like you’re playing a cousin of the original Mario Bros.

And though this initial version was built as a tool for kids with autism, the dialogues are all customizable. In fact, the whole game is completely open source, creating an opportunity for multiple applications. Anderson said others who’ve looked at it see potential for using the game with kids with other learning or physical disabilities, those who have been bullied and retreat into an online social world, or even with seniors vulnerable to online scams.

The game is not only a realization of a long-held dream for Anderson, but a major accomplishment for the students who built it.

At the recent CECS senior design competition, the project took home the Computer and Information Science Department’s first place prize and tied for most innovative.

When asked about their success, the team members earnestly credit “every year and every class” of their education—which they said gave them the huge array of skills you need to do something as audacious as building a video game from scratch.

“It kind of felt like it took four or five years of experience and mashed it all into one project,” Papaioannou said.

“And I think we succeeded because we didn’t underestimate it,” Croskey added. “People think you just start with a game engine, you put in the art, and everything else is done for you. But really, almost everything in there—from what happens when a character falls to all the database work behind the scenes—we did ourselves.”

He’s not exaggerating. Team member Dominic Retli even wrote the original music for the game; and he and Papaioannou hand drew many of the characters to supplement the game artwork donated by Chicago-based artist Alex Van Trejo. They didn’t get any classroom training for that.

But that’s the kind of pluck it takes to finish the unfinishable.

Experiential Learning