Data Science newsletter – March 1, 2022

The city of Orlando is investing in a digital twin of its metro region, which will use real-time data to inform decisions on infrastructure, utilities and business development.

Unlike many other cities’ digital twins that capture downtown areas, the Orlando project — currently being developed by technology company Unity — will map the entire 40-square-mile metro region, including the Space Coast, amusement parks and the Orlando International Airport. That will allow it to also be used as a model to show off the region to businesses and developers interested in investing in the city, said Orlando Economic Partnership (OEP) CEO Tim Giuliani.

The model, slated to be unveiled in October, is a key part of the city’s “digital transformation,” Giuliani said

Many of today’s AI systems are designed around the big red button, he says. “Whatever we want from AI, we’re going to press the button, and as if by magic, it’s going to deliver the result to us,” says [Ge] Wang. The perception is that “AI has this magical quality, in the sense that it exhibits this, for lack of a better word, intelligence. And it’s able to do complex tasks. AI is the most powerful pattern-recognizer that we’ve ever built.”

The question becomes, then, ” what do we really want from AI?” Wang continues. “Do we want oracles all the time, that just give us the right answers without showing its work necessarily? Or, do we want tools? Things that we can use to learn to get better at? And tools that are, by definition, interactive to the human?”

The ways to bring humans more into the AI loop don’t necessarily have to be complicated.

Engineers at UC Berkeley have developed a new technique for making wearable sensors that enables medical researchers to prototype test new designs much faster and at a far lower cost than existing methods.

The new technique replaces photolithography — a multistep process used to make computer chips in clean rooms — with a $200 vinyl cutter. The novel approach slashes the time to make small batches of sensors by nearly 90% while cutting costs by almost 75%, said Renxiao Xu (Ph.D.’20 ME), who developed the technique while pursuing his Ph.D. in mechanical engineering at Berkeley.

“Most researchers working on medical devices have no background in photolithography,” Xu said. “Our method makes it easy and inexpensive for them to change their sensor design on a computer and then send the file to the vinyl cutter to make.”

A group of researchers at MIT, in collaboration with researchers at Harvard University and Fujitsu Ltd., sought to understand when and how a machine-learning model is capable of overcoming this kind of dataset bias. They used an approach from neuroscience to study how training data affects whether an artificial neural network can learn to recognize objects it has not seen before. A neural network is a machine-learning model that mimics the human brain in the way it contains layers of interconnected nodes, or “neurons,” that process data.

The new results show that diversity in training data has a major influence on whether a neural network is able to overcome bias, but at the same time dataset diversity can degrade the network’s performance. They also show that how a neural network is trained, and the specific types of neurons that emerge during the training process, can play a major role in whether it is able to overcome a biased dataset.

Identifying a malfunction in the nation’s power grid can be like trying to find a needle in an enormous haystack. Hundreds of thousands of interrelated sensors spread across the U.S. capture data on electric current, voltage, and other critical information in real time, often taking multiple recordings per second.

Researchers at the MIT-IBM Watson AI Lab have devised a computationally efficient method that can automatically pinpoint anomalies in those data streams in real time. They demonstrated that their artificial intelligence method, which learns to model the interconnectedness of the power grid, is much better at detecting these glitches than some other popular techniques.

Because the machine-learning model they developed does not require annotated data on power grid anomalies for training, it would be easier to apply in real-world situations where high-quality, labeled datasets are often hard to come by.

We write to announce that Yale is creating endowment funds totaling $250 million to support the development of future leaders in medicine, nursing, and public health. For the past two years, the pandemic has placed enormous strain on the workforce in these fields and increased the need for public health experts and health care professionals worldwide. Health professionals—including our colleagues from the Yale School of Medicine (YSM), School of Nursing (YSN), and School of Public Health (YSPH)—have worked with great fortitude and dedication to save lives; develop COVID vaccines, treatments, and tests; and inform health policy. Their expertise will be even more critical as we recover from this pandemic and prepare to face adverse health effects from new and existing infectious and chronic diseases, health care inequities, resource scarcity, and other pressing challenges. Yale is committed to help address global demand for leaders who can guide individuals and communities to better health.

A First Try at ROI: Ranking 4,500 Colleges finds that bachelor’s degrees from private colleges, on average, have higher ROI than degrees from public colleges 40 years after enrollment. Community colleges and many certificate programs have the highest returns in the short term, 10 years after enrollment, though returns from bachelor’s degrees eventually overtake those of most two-year credentials.

ROI of Liberal Arts Colleges: Value Adds Up Over Time finds that the median ROI of liberal arts colleges is nearly $200,000 higher than the median for all colleges. Further, the 40-year median ROI of liberal arts institutions ($918,000) is close to those of four-year engineering and technology-related schools ($917,000), and four-year business and management schools ($913,000).

When the research arm of the U.S. Department of Education wanted to learn more about the latest advances in robo-grading, it decided to hold a competition. In the fall of 2021, 23 teams, many of them Ph.D. computer scientists from universities and corporate research laboratories, competed to see who could build the best automatic scoring model.

One of the six finalists was a team of just two 2021 graduates from the Georgia Institute of Technology. Prathic Sundararajan, 21, and Suraj Rajendran, 22, met during an introductory biomedical engineering class freshman year and had studied artificial intelligence. To ward off boredom and isolation during the pandemic, they entered a half dozen hackathons and competitions, using their knowhow in machine learning to solve problems in prisons, medicine and auto sales. They kept winning.

“We hadn’t done anything in the space of education,” said Sundararajan, who noticed an education competition on the Challenge.Gov website. “And we’ve all suffered through SATs and those standardized tests. So we were like, Okay, this will be fun. We’ll see what’s under the hood, how do they actually do it on the other side?”

[William] Chueh, lead author Haitao “Dean” Deng, PhD ’21, and collaborators at Lawrence Berkeley National Laboratory, MIT and other research institutions used artificial intelligence to analyze new kinds of atomic-scale microscopic images to understand exactly why batteries wear out. Eventually, they say, the revelations could lead to batteries that last much longer than today’s. Specifically, they looked at a particular type of lithium-ion batteries based on so-called LFP materials, which could lead to mass-market electric vehicles because it does not use chemicals with constrained supply chains.

The University of Florida announced this week a new collaboration with tech giant IBM (NYSE: IBM) to launch a comprehensive skills program designed to extend UF’s vision to be an international leader in artificial intelligence, data science, fintech, and other related technologies that can help solve society’s biggest challenges.

UF — already ranked by U.S. News & World Report as one of nation’s most innovative universities — and IBM will work together to support UF’s faculty and students as they develop diverse and high-demand skillsets in artificial intelligence, cybersecurity, quantum cloud computing and data science that align with industry needs and trends. The collaboration will extend to West Palm Beach, where UF is exploring an opportunity to co-create academic programming at a new campus that will serve the needs of the region’s rapid influx of companies across sectors, including finance and technology.

The collaboration is the next step in UF’s ambitious goal to be the leading “AI University” in the nation. The initiative will help UF transform the nation’s workforce and bolster research by embedding technology into its curriculum across disciplines.

The adoption of RISC-V, a free and open-source computer instruction set architecture first introduced in 2010, is taking off like a rocket. And much of the fuel for this rocket is coming from demand for AI and machine learning. According to the research firm Semico, the number of chips that include at least some RISC-V technology will grow 73.6 percent per year to 2027, when there will be some 25 billion AI chips produced, accounting for US $291 billion in revenue.

The increase from what was still an upstart idea just a few years ago to today is impressive, but for AI it also represents something of a sea change, says Dave Ditzel, whose company Esperanto Technologies has created the first high-performance RISC-V AI processor intended to compete against powerful GPUs in AI-recommendation systems. According to Ditzel, during the early mania for machine learning and AI, people assumed general-purpose computer architectures—x86 and Arm—would never keep up with GPUs and more purpose-built accelerator architectures.

Almost half of the US population will interact with conversational AI in 2022, through the use of chatbots and virtual assistants.

Conversational AI, in the form of chatbots or virtual assistants, will be used by 123 million US adults at least once a month, or 47 percent of the population.

That’s according to new research by eMarketer, which predicts that the technology will be used by 130 million US adults by 2025, or about half the population.

Over the last decade or so, international studies on human population genetics have begun to expand genomic libraries to encompass regions of the Global South — including Southeast Asia, where I am a science reporter, and the Pacific islands. These international studies, often led by Western scientists, have contributed to a more global understanding of ancient patterns of human migration and evolution. But on some occasions, they’ve also sidestepped local regulatory agencies in the developing world, and ventured into murky research ethics terrain as a result.

A recent example — a case that simultaneously illustrates the promise, pitfalls, and pressure points of international genomics research — comes from the largest genetic study ever conducted in the Philippines, published last year in the Proceedings of the National Academy of Sciences. A team led by Mattias Jakobsson of Uppsala University in Sweden and Maximilian Larena, who was a researcher in Jakobsson’s lab at the time, collected and analyzed DNA samples from more than 1,000 Filipinos representing 115 Indigenous groups. The study determined that today’s Filipino population descends from at least five distinct waves of human migration, spanning thousands of years — a finding that they said contradicted the prevailing theory of how humans populated the islands.

Schar School of Policy and Government professor J.P. Singh leads a team of researchers from across George Mason University campuses that has been awarded a three-year, $1.39 million grant to study the economic and cultural determinants for global artificial intelligence (AI) infrastructures—and describe their implications for national and international security.

The grant was awarded by the Department of Defense’s esteemed Minerva Research Initiative, a joint program of the Office of Basic Research and the Office of Policy that supports social science research focused on expanding basic understanding of security. This year’s 17 university teams were selected from more than 220 proposals and will receive a total of $28.7 million in grants.

“Perhaps the most important and unique feature of our project is that we will use sophisticated big data models in computer and social sciences to specify the economic and cultural determinants,” said Singh, the principal investigator of the grant.

The twenty-two-year-old figure skater Nathan Chen grew up in Salt Lake City, where he took his first lessons wearing his older sister’s hand-me-down skates, on a practice rink for the 2002 Olympic Games. By the age of ten, he’d won the novice title at the national championships. His mother, a Chinese immigrant, began driving him to California to train; to save cash, they sometimes slept in the car instead of booking a hotel room. At his Olympic début, in 2018, in Pyeongchang, South Korea, Chen was a favorite to win gold, but he fell during his short program and entered the free skate in seventeenth place. With little to lose, Chen ended those Olympics on a note of go-for-broke glory, landing five planned quadruple jumps in his long program and, just for kicks, adding in an unexpected sixth. The performance wasn’t enough to get him on the podium—he ended up finishing fifth—but it lingered in the minds of fans as a display of his virtuosic talent. He is often called the Quad King.

In 2018, Chen enrolled at Yale University, where he studies statistics and data science. Before taking a leave of absence to train for his second Olympics, he’d spend the mornings in class and save the afternoons for skating at the campus rink. Earlier this month, in Beijing, Chen earned the gold medal that had eluded him in the men’s singles, topping off a record-breaking short program with a soaring free skate, scored to an Elton John medley.

We, a subset of Computer Science juniors, hope to bring to the college’s attention the issues we have faced in the major, and how these issues have a unifying cause: the understaffing of the Reed CS department. Our most pertinent concerns are with our upcoming thesis experience, our experience with grading and feedback of courses in the department, the inconsistency of course material and syllabi for required courses in the major, mishaps regarding the critical Junior Qualifying Exam, and how all of these connect back to the understaffing of our unsupported department.

The most direct consequence will be on our upcoming thesis. There are about 30 rising seniors ready to start thesising, with 2 CS professors in the worst case. This means that there will be roughly 15 theses per professor. This is an unprecedented and unreasonable thesis-faculty ratio. The professors simply won’t be able to provide the assistance and support to thesising seniors. We juniors are fearful that our thesis experience will be altered or removed altogether.