In the case of Priya Donti, childhood trips to India were more than an opportunity to visit a further family. The Travel Biennale was activated in it by motivation, which still shapes her research and teaching.
Contrasting her family home in Massachusetts, Donti – now a professor of Silverman’s family career development at the Faculty of Electrical Engineering and Computer Science (EECS), a common position between the myth of Schwarzman College of Computing and ECS, and the main investigators in MIT laboratory for information and decision systems (LIDS).
“It was very clear to me, to what extent unevenness is a raging problem all over the world,” says Donti. “From an early age I knew that I definitely wanted to solve this problem.”
This motivation was additionally excited by a biology teacher at high school, who focused his class on the climate and sustainable development.
“We learned that climate change, this huge, important issue, would exacerbate unevenness,” says Donti. “It really got stuck with me and the fire in my stomach.”
So, when Donti enrolled in Harvey Mudd College, she thought she would direct her energy towards learning about chemistry or material materials to create a recent generation solar panels.
However, these plans were abandoned. Donti “fell in love” in computer science, and then discovered the work of scientists from Great Britain, who argued that artificial intelligence and machine learning would be necessary to lend a hand integrate renewable energy sources.
“For the first time I saw these two interests,” he says. “Since then I have become addicted and from now on I am working on this topic.”
Doti, when implementing a doctorate at the Carnegie Mellon University, was able to design her diploma to include computer science and public policy. In her research, she examined the need for basic algorithms and tools that could be managed on a immense extent on renewable energy sources on a scale.
“I wanted to develop these algorithms and sets of tools, creating new machine learning techniques based on computer science,” he says. “But I wanted to make sure that the way I did the work was based on both the domain of energy systems and working with people in this field” to ensure what was needed.
While Donti worked on her doctorate, she co-founded a non-profit organization called AI Climate Change. She says that its goal was to lend a hand the community of people involved in climate and sustainable development – “are they computer scientists, scientists, practitioners or decision -makers” – in combination and access to resources, merger and education “to help them on this journey”.
“In climate space,” he says, “you need experts in particular sectors related to climate change, experts in various sets of technical tools and social sciences, problems of problems, affected users, decision-makers who know the regulations-all of them-to have a scalable influence.”
When Donti came to the myth in September 2023, it is not surprising that her initiatives attracted her to the exploit of computer science to the biggest problems of society, especially the current threat to the planet’s health.
“We really wonder where technology has a much longer influence and how technology, society and politics must cooperate,” says Donti. “Technology is not only one and monetized in the context of the year.”
Her work uses deep learning models to include physics and challenging limitations of power systems that exploit renewable energy sources to better forecast, optimize and control.
“Machine learning is already really widely used for such things as forecasting solar energy, which is a condition for managing and balancing energy networks,” he says. “I focus on how you improve the algorithms of the actual balancing of the power network in the face of many changing energy sources?”
Donti breakthroughs include a promising solution for power network operators to be able to optimize costs, taking into account the actual physical reality of the grid, instead of relying on approximations. Although the solution has not yet been implemented, it seems that it works 10 times faster and much affordable than previous technologies and attracted the attention of network operators.
Another technology that it develops works to provide data that can be used in training machine learning systems to optimize the power system. In general, a lot of data related to systems is private because they are reserved or because of security problems. Donti and its research group are working on creating synthetic data and comparative tests, which, according to Donti, “can help reveal some of the basic problems” in increasing the efficiency of power systems.
“The question is,” says Donti, “Can we bring our data sets to some extent that they were difficult enough to increase progress?”
The American Department of Computational Science Graduate Fellowship and NSF Graduate Research Fellowship received Donti for her efforts. It was recognized as part of the list “35 innovators below 35” and VOX “Future Perfect 50”.
The next spring, Donti will be a co -inspected AI class on climate activities with Sara Beery, an assistant professor, whose goal is artificial intelligence for biological diversity and ecosystems, as well as Abigail Bodner, assistant to the professor in EEC and Earth departments, atmospheric and planetary, which are focused on air conditioning and earth.
“We are all very excited,” says Donti.
Arriving in the myth, Donti says: “I knew that there would be an ecosystem of people who really care, not only success indicators, such as publications and the number of citations, but about the impact of our work on society.”