Machine Learning – Frost Institute for Data Science & Computing

November 16, 2023February 27, 2024

Five years ago, representatives from telecommunications companies approached Yelena Yesha in a panic. As they sent more and more satellites into orbit around the Earth, the potential risks for hacking grew with each launch and they wanted guidance from Yesha, a world-renowned computer scientist who has worked extensively in cybersecurity.

In the years since, corporate fears started becoming reality. In 2022, a Belgium researcher hacked one of Elon Musk’s Starlink nanosatellites by rigging a ground-based user terminal with $25 worth of off-the-shelf parts. In 2023, a multinational team of researchers proved it could hack into a satellite owned by the European Space Agency, and a leaked CIA report revealed that China is building cyber weapons to ‘‘seize control’’ of American satellites.

As cyber incidents become common with thousands of satellites now circling the Earth, Yesha was approached again this year, but by a different group: NASA. The government space agency was looking for ways to protect its growing constellation of satellites and signed a contract with Yesha’s team at the University of Miami’s Frost Institute for Data Science and Computing (IDSC) to help mitigate cyber risks.

“Anybody can mess with the data, corrupt it,
put something that doesn’t belong there”

Yesha, who runs the AI +Machine Learning unit at IDSC, described the response as a monumental task, especially to address concerns for smaller craft (commonly referred to as nanosatellites or microsatellites), with limited hardware, software, and power budgets. And given that satellites mostly communicate with the ground through basic radio signals, or network between satellites, ‘‘anybody can mess with the data, corrupt it, put something that doesn’t belong there’’ Yesha said. ‘‘We need to protect intersatellite intercommunication in space.’’

Yesha’s team, including Research Associate Professor Phuong Nguyen, Professor of Professional Practice Stephen Dennis, and Assistant Scientist Yusen Wu, are developing a new computational platform that will be used to explore the security of satellite operations using a simulation environment, said Jacqueline Le Moigne, Advanced Information Systems Technology (AIST) Program Manager for NASA’s Earth Science Technology Office (ESTO).

“Dr. Yesha’s work will provide NASA with
zero-trust cybersecurity technologies”

‘‘NASA is developing revolutionary capabilities in terms of Intelligent and Collaborative Constellations of Smallsats communicating seamlessly with ground and airborne sensors,’’ Le Moigne said. ‘‘Dr. Yesha’s work will provide NASA with zero trust cybersecurity technologies that will ensure that all of these various observing nodes will be able to securely communicate and interact to form the future ‘Internet of Earth-Things’.”

The core of the new system that Yesha’s team will develop will rely on blockchain technology. The general public associates that technology mostly with cryptocurrencies, which have suffered several high-profile collapses in recent years. But Yesha says the underlying technology will not only endure but become more prevalent across all kinds of systems. ‘‘Blockchain will power 10 percent of the entire economy,’’ she said.

Blockchain architecture can be permissioned for controlled information sharing with an immutable ledger that facilitates transaction audits, making it easy for regulatory stakeholders to detect and track illegitimate access. The technology may also streamline the way in which governments and private companies allocate and manage satellite time.

“Data stored in ledgers provides significant protection, making it
much more difficult for adversaries to gain illicit access”

Some private companies, like Planet Labs and Capella Space, allow customers to use their basic, web-based system to order satellite imagery on demand, but that leaves room for hackers to step in. Yesha says blockchain technology will make that process more secure and efficient since every person who accesses each satellite will be tracked and recorded. The technology will also help secure all the data collected by satellites since the owners of the satellite will be able to see if anybody accessed or changed the data. ‘‘This data stored in ledgers provides significant protection making it much more difficult for adversaries to gain illicit access,’’ according to a summary of Yesha’s project.

The end result, Yesha hopes, will help to guide the development of secure satellite systems that can be used by commercial companies and government agencies, including those operated by the National Oceanic and Atmospheric Administration (NOAA), the Department of Defense, and the emerging Space Force.

Boosting Product Development

January 16, 2023October 24, 2024

When Yelena Yesha, PhD, was hired at the Frost Institute for Data Science and Computing (IDSC) three years ago, part of her charge was to take the groundbreaking research being done at the University of Miami and translate it into private, income-generating companies.

Now, the world-renowned computer scientist has teamed up with a chemical formulation and product design expert at the College of Engineering to launch a startup company they believe will revolutionize the way multinational companies develop their products.

Yesha partnered with Dr. Samiul Amin, Professor of Practice at UM’s Chemical, Environmental, and Materials Engineering Department, to form FastFormulator, a company that uses Artificial Intelligence and Machine Learning (AI/ML) to fast-track the development of compounds used in cosmetics, cleaning products, biopharma, foods, and inks. The duo is trying to replace the traditional, time-consuming process of chemical formulation and product development that still relies on old-fashioned trial and error with a predictive approach that uses AI/ML to quickly create new compounds that do exactly what they’re supposed to do.

Yesha said the company represents the kind of commercial research spinoff she is trying to foster at IDSC and throughout the UM campus. “This is very much in the realm of my vision and why I came to UM,” said Yesha, who also serves as the Knight Foundation Endowed Chair of Data Science and AI. “There are a lot of great ideas on campus that have this potential. I would like to dedicate more of my time, energy, and know-how to these types of efforts, and see how we can help faculty and students around campus realize their dreams.”

FastFormulator came about nearly by chance when Amin and Yesha met at an event on UM’s Coral Gables campus in January of 2023. Amin had spent more than two decades working for some of the biggest chemical companies in the world, including Loreal, Unilever, and Exxon/Mobil. He left the private sector and joined UM as a Professor of Practice in the hopes of finding computer scientists who could help with his idea of using AI/ML to streamline the chemical formulation and product design process.

Yesha had spent her career developing AI/ML techniques that have been used to create electronic commerce systems, repair the Hubble Space Telescope, identify lung cancers, and mine health care data to forecast illness and create new therapies, but had no experience with chemical development.

Yet the two bonded immediately and saw an opportunity to upend the way companies (cosmetics, consumer goods, foods, coatings, etc.) develop their formulated products. “Yelena saw it quickly,” Amin said. “We hit it off.”

Amin said the formulated-products-based companies such as cosmetics and consumer goods are facing pressure from consumers and regulators to completely reengineer their products so they’re more eco-friendly and sustainable, but are struggling to do so because their development processes are so slow.

Amin uses the example of shampoos to illustrate the point. Most shampoo companies developed their signature products decades ago, only tinkering with the complex formulas from time to time. Now they’re being forced to remove problematic ingredients, replace them with more eco-friendly ingredients, and still deliver the same product. But since the companies still use experiment-intensive development techniques, making such wholesale changes has been slow, costly, and ineffective.

“I can give you the most green shampoo on earth, the most biodegradable, but it doesn’t clean your hair, it doesn’t foam up, and it’s water-thin. You’re never going to sell that,” he said.

FastFormulator is designed to make broad changes quickly while maintaining the quality and feel of the original products. Rather than testing out each chemical combination one by one, Yesha and Amin have started cataloging the effects of each ingredient (and untold combinations of those ingredients) and training a computer to quickly test out thousands of combinations to identify the ideal formula.

“AI/ML will actually identify those gold nuggets in the formulation, where other people, just by the sheer number of experiments you need to get there, may never know exist.”

“AI/ML will actually identify those gold nuggets in the formulation, where other people, just by the sheer number of experiments you need to get there, may never know exist,” Amin said.

The duo fast-tracked their research and, within one year, have already patented the concept and incorporated their new business. Since their research was developed at the University of Miami, they had to negotiate with the University’s Office of Technology Transfer (OTT), which oversees the commercialization of all innovations developed on campus. The OTT obtained a patent on the process developed by Amin and Yesha, allowing the duo to co-found FastFormulator with UM retaining equity in the company and a percentage of profits.

The company now has a standalone office/laboratory at Converge Miami, has signed several clients and hired two full-time contractors who are already working on the company’s first projects. Yesha said they will soon hire a Board of Advisors and a CEO to run the day-to-day operations, as Yesha and Amin continue fine-tuning the product and recruiting more clients.

“We want the community to know
that these things are possible at UM.”

Yesha intends to spend about one day a week focused on the new company, but the rest of her time will be spent back at UM searching for the next big opportunity. “We want the community to know that these things are possible at UM,” she said.

Story by Alan R. Gomez