Penn State will hold the first National Science Foundation-funded symposium on how machine learning can contribute to the field of hydrology. Credit: iStock/Design Cells. All Rights Reserved.
Penn State to hold NSF-funded machine learning symposium on water resources
Posted on March 30, 2022UNIVERSITY PARK, Pa. — Penn State will hold the first National Science Foundation-funded symposium on how machine learning can contribute to the field of hydrology. The event, titled “HydroML Symposium on Big Data Machine Learning in Hydrology and Water Resources,” will be held May 18-20 in person at University Park and online. Early registration ends on April 1, while regular registration ends on May 1.
“This symposium emphasizes building a community that has common goals and shared resources,” said Chaopeng Shen, associate professor of civil and environmental engineering and chair of the organizing committee. “We are attempting to demystify machine learning for those new to the field, build up machine learning expertise, encourage collaboration among those already involved and build a machine learning community for all participants.”
From where water exists and how it moves to its quality and how it’s used, the study of water can impact every aspect of life, according to Shen. Shen has previously applied artificial intelligence technologies to model and predict landslides and to forecast near-real-time soil moisture — a critical indicator of locust breeding capabilities — during the 2020 infestation in East Africa, among several other projects. One such technology is called deep learning, a machine learning technique that learns to recognize certain features from data and how to process the information based on related tasks.
“Deep learning is a tool that we can leverage to make predictions better, more accurate, and less expensive — with less effort,” said Shen, who is also affiliated with the Penn State Institute for Computational and Data Sciences, which is also supporting the symposium. “AI tools like machine learning can help us better understand how parameters fit together and influence forecast performance. More recently, the integration between AI and physics is leading to promising new avenues, and it also attracts more people. Yet no specific conferences exist to build this community.”
Three years ago, machine learning was already developing rapidly in the sub-field of AI as applied in hydrology, according to Shen, but there was a lack of communication between the people using it.
“There wasn’t a conference or meeting focused in this area, where we could come together to share findings or approaches and make the networking connections needed to move research forward,” Shen said. “There are so many people working on so many topics in this field, and nothing to pull us together, so I applied for a grant from NSF to fund the first symposium in this area.”
The grant was awarded in early 2020, with plans for a meeting that year. Due to the pandemic, the symposium was delayed to this year, with lifted COVID-19 restrictions. The organizing committee will continue to monitor COVID-19 case levels and comply with all federal, state and University guidelines pertaining to the virus. If circumstances change, symposium registrants will be notified.
“The people who are using machine learning to better understand hydrology have not had the opportunity to discuss what the future of the field looks like at a high level,” Shen said. “That type of conversation is difficult to have virtually, especially when you’re talking about building a community of different people doing different things who don’t need to work closely together on their specific projects. So, we waited until we could gather in person.”
The symposium will include plenary and lightning talks, poster presentations, machine learning tutorials, hackathons, networking opportunities, and social events. The discussed research topics will cover all sub-areas of hydrology and a variety of machine learning approaches, Shen said.
“The symposium participants have a shared mission: We think AI is going to, first, improve our predictive capabilities; second, improve our understanding; and third, answer many standing questions currently without solutions,” Shen said. “I hope that it will bring researchers together so that we can strengthen both our individual efforts and the field as a whole.”
For more information, including how to register and details on the organizing committee, please visit the HydroML website or email the organizing committee.