These days, it’s all about the data,” said Carola Wenk, professor of computer science at Tulane. “Lately, there’s been a data explosion.” In every field you can name, data has proliferated in recent decades. Computer scientists — and everyone working with data — can gather, manipulate and store it in larger quantities than ever before.

Computer scientists at Tulane are exploring and discovering new roles for data in everything from GPS technology, to cancer diagnoses to fake news — even to taking the dog for a walk. And during the COVID-19 pandemic, data about coronavirus testing, cases and deaths has been essential to public health efforts to combat and understand the disease.

“Everybody has data and wants to do something with it,” said Wenk. “And we now have processing power to deal with a lot of data.”

Computer scientists like Wenk and other colleagues at Tulane such as Jihun Hamm, associate professor of computer science, and Nicholas Mattei, assistant professor of computer science, are diving deep into artificial intelligence.

They are exploring deep neural networks and deep learning to address all kinds of problems.

Computational Problem Solving

By training, Wenk is a mathematician. She earned a PhD from the Free University of Berlin in 2002 and has received numerous research grants and awards from the National Science Foundation, the National Institutes of Health and the Defense Advanced Research Projects Agency (DARPA). She works in computational geometry in the area of algorithms.

Algorithms are step-by-step instructions. It’s how “you tell a computer what to do,” Wenk said. And you do that by feeding it data. Like a recipe for a tasty dish depends on fresh ingredients, algorithms rely on new data.

Wenk is interested in the potential of computer science beyond machine learning, per se. What drives her is interdisciplinary work and collaboration.

“It’s all about computational problem solving,” she said. “You want to solve the problem, and there are different tools out there. The computer scientist needs to work with domain experts on developing solutions together.”

Wenk has collaborated with domain experts such as physicians, geoscientists, ecological biologists and social scientists.

Much of her work is in imaging data, and she’s contributed to tracking and understanding the movement of cars and seagulls.

Her breakthrough research includes the refined definition and measurement of the Fréchet Distance, which is used to compare shapes of curves. An everyday explanation of the Fréchet Distance includes man’s best friend. If a man is traveling on a curved path while his dog is on a leash traveling a separate curve, the Fréchet distance between the two curves is the length of the shortest leash sufficient for both man and dog to travel their separate paths.

Utilizing data from satellites, the Fréchet Distance is useful for analyzing Global Positioning System trajectories of cars on roads.

Wenk is also collaborating on a project with J. Quincy Brown, associate professor of biomedical engineering, and Brian Summa, assistant professor of computer science, that is far removed from cars or birds or walking dogs. Brown, the Paul H. and Donna D. Flower Early Career Professor in Engineering, is developing new diagnostic procedures for prostate cancer. With her data imaging expertise, Wenk and Summa are helping to evaluate the visual patterns of the biopsies of men with prostate cancer, detecting whether they have fast- or slow-growing cancer.

“You want to decide between, is this person healthy, or is this person sick?” Does the patient need invasive surgery, or will less intrusive treatment be effective?

Wenk has also branched out to working with social scientists on a DARPA “ground truth” project to understand movement patterns of humans and social networks. “With more computational power, it’s feasible to have the computer learn a bit more similar to what a human does,” said Wenk.

The Renaissance of AI

Jihun Hamm, associate professor of computer science, agrees that, with increased power, computers are learning more and more in the same ways that humans learn.

“A lot of the human brain is devoted to the vision process,” said Hamm. “And now computers are able to recognize, or are getting very close, to human-level visual recognition.”

Hamm received his PhD from the University of Pennsylvania in 2008. His specialty is machine learning, a subfield of computer science. “I’m teaching machines to learn,” he said. “And I can teach people to learn how to make machines learn.”

Artificial intelligence (AI) had stalled since the 1980s and entered a period of dormancy. “It was the winter of AI,” said Hamm. “People had high expectations of what AI can do, and the reality couldn’t catch up with the expectations.”

But in the 21st century, especially during the past five or six years, artificial intelligence is experiencing a renaissance, said Hamm.

“Machine learning became a new savior for AI,” Hamm said. “With progress in machine learning, combined with progress in hardware development, big strides in AI were possible.”

Among the latest developments and pitfalls are algorithms that can be used to generate data that looks real to the human eye but is actually synthesized. “It can be difficult for a human to distinguish synthesized images from real images,” said Hamm. As a result of this new technology, fake videos and audio may be produced.

“It’s hard to know where we will go from here,” said Hamm. “But the level of realism in the data generated by machines, such as images, photos and all this, is quite astonishing.”

A core principle of machine learning is learning theory, which involves mathematical proofs. “I believe computer science is a branch of applied mathematics,” said Hamm. “And I don’t feel comfortable claiming something without proof.”

Exploring concepts of adversarial learning in which two or more agents compete with each other to generate realistic data to find an optimal state of an algorithm “is my way of trying to make this machine learning problem concrete and rigorous,” said Hamm, who is bringing on board postdoctoral and doctoral researchers to his lab.

“I want to have something that is clearly explainable, why something works and why something doesn’t work.”

Ethical Issues in AI

Access to data is “arguably what computer science people have wanted since the beginning,” said Nicholas Mattei, assistant professor of computer science, “because data is the thing that drives how we build and test these systems.”

Mattei teaches data science and artificial intelligence at Tulane. He earned his PhD in computer science from the University of Kentucky in 2012.

Systems are built, using data, to predict human behavior, whether it’s what a person will want to watch next on Netflix or purchase from Amazon, or how they will act in more complicated ways.

“The way we collect all this data now is so easy and almost transparent to most people,” said Mattei. “Technology intermediates many, many things we do.”

For example, Google Maps can conveniently alert drivers to congestion on the road ahead. That’s possible because other drivers have given up data about slowdowns they are encountering when they are connected to Google Maps, too.

“We all sort of gain,” said Mattei. But “sometimes, I think a big problem is that people are giving up their data in ways they’re not sure of. We have trouble understanding the scope of data collected and how that data can be used.”

Mattei is writing a book, Understanding Technology Ethics Through Science Fiction, under contract with MIT Press, with four co-authors. It’s about technology’s impact on our lives.

“A lot of the impacts that people talk about as being AI or computer science are a confluence of many different advances in multiple fields of technology,” he said.

The goal of the book, which is geared toward computer science students and others interested in broader technology issues, is to get readers “to think about these things — not in prescriptive terms but to give them the tools and language of ethical theory and descriptive ethics.”

Topics addressed in the book are technology’s “ramifications for privacy and how we look at interpersonal relationships and knowledge.”

The notion of legal privacy as “information about you” to be protected is relatively new in human history, said Mattei.

It came about in the early 1900s after new technologies, such as the phonograph, microphone and telephone, were invented and came into widespread use. These advances allowed people to be recorded and for those recordings to be moved and reproduced without their knowledge.

Mattei’s book will offer essays by computer scientists and religious studies and ethics scholars. To spice up the book, science fiction stories also will be included.

One story is “Here-And-Now” by Ken Liu. It explores privacy issues as a young man navigating the gig economy inadvertently discovers information about his parents’ marriage when he connects with a work-for-hire app to earn extra money. Stories like this add to the conversation about the influence of technology on society.

“There’s lots of room for creativity in computer science,” said Mattei. And computer science is a place with great opportunity for new graduates.

Not a Typical Program

At Tulane, computer science is a coordinate major, which means students study in other disciplines such as business, public health and liberal arts as well as science and engineering.

In a typical engineering program — actually “in all of the technology space, writ large in engineering” said Mattei — “you learn how to do things. But maybe they don’t teach you how to think about what you can do.”

At Tulane, sociological, philosophical and economic issues are not separate from computer science because it is a cross-disciplinary program.

Hamm is teaching machine learning, a “hot” course with students eager to enroll. He said the undergraduate and graduate students he has taught at Tulane are “frank and open-minded on discussing their problems and discussing their progress on the projects. I find interacting with the students very engaging.”

Wenk has taught and advised students with primary majors in business, biomedical engineering, engineering physics, mathematics, linguistics and even a music major who worked on a computational musical art installation as his computer science capstone project.

Wenk has been at Tulane since 2012, when the computer science department was reestablished. “We built this department around this collaborative, interdisciplinary theme,” she said.

Another unusual aspect of the department is “we have more female undergraduates in our program. We have 30 to 40% women, which is unheard of for computer science.

“We have a different atmosphere in class,” she said. “The students have a different mindset.”

With all the precautions in place to slow the spread of the coronavirus, “We have embraced technology to collaborate and teach online, such as using collaborative coding platforms or collaborative whiteboards to solve problems together in groups or whole classes,” Wenk said.

Wenk even presented lectures for her algorithms graduate course outdoors on campus under a tree and in Audubon Park.

“Computer science students and faculty have done a remarkable job adapting to the new socially distanced teaching and research environments.”

Additional Data Science Programs at Tulane

Master of Business Analytics at the A. B. Freeman School of Business
The mission of the Master of Business Analytics program is to prepare its graduates for careers that allow them to manage and make data-driven decisions. The program is a STEM–designated, 10-month full-time graduate program designed for recent graduates and those with 1-2 years of work experience.

Master’s Degree in Computational Science at the School of Science and Engineering
The Center for Computational Science offers a master’s degree in computational science in combination with a student’s B.S. degree in science or engineering.  This is known as a “4+1” program.  Tulane undergraduate students can earn the master’s degree along with their bachelor’s degree in just one extra year at reduced tuition. All engineering and science majors can apply.

Master of Science in Biomedical Informatics at the School of Medicine
The goal of this two-year (four-semester) thesis program is to train new Biomedical Informatics (BMI) specialists. These uniquely trained master’s graduates will be critical to existing efforts to improve health outcomes. The program prepares graduates to participate in research programs in academia, health care, public health and industry, as well as to apply the knowledge in clinical, government and industry settings.