Oct. 5 researcher panel to discuss AI for predictive science

UNIVERSITY PARK, Pa. — Artificial intelligence is affecting society worldwide, from improving medical diagnostic tools, to bolstering supply chains and refining weather forecasting, along with many more applications. Closer to home, Penn State researchers are also leveraging AI techniques in their work to advance their science.

The Institute for Computational and Data Sciences (ICDS), which supports Penn State researchers using advanced research computing techniques, including AI, is bringing together a group of Penn State researchers to discuss their work using AI for predictive modeling and analytics, as well as the ethical considerations of AI. The panel, “AI for Predictive Science,” will be held online at 2 p.m. on Monday, Oct. 5. The event is free and open to the public, and advance registration is required.

“Predictive science methods and solutions are generally ensconced in existing algorithms and software based on sound physics, mathematics and statistics. However, they often lag behind the revolution in computational power and artificial intelligence and machine learning. There is a potential transformative solution to this problem by leveraging recent advances in big data analytics,” said Guido Cervone, professor of geography and meteorology and atmospheric science, and ICDS associate director, who will moderate the panel. “This panel will discuss AI/ML within the context of making future predictions, giving the perspective of several colleagues who are active in this field. Furthermore, it will give the opportunity to participate in ICDS activity to catalyze resources and expertise for future initiatives.”

The panel will include four researchers: David Hughes, associate professor of entomology and biology; David Hunter, professor of statistics; Chaopeng Shen, associate professor of civil and environmental engineering; and Jian Sun, postdoctoral scholar in the College of Earth and Mineral Sciences. Hunter plans to discuss the ethical considerations of AI by sharing his experience related to a court case that went before the U.S. Supreme Court. The other panelists will discuss their work with developing and using AI tools.

Hughes is the founder of PlantVillage, an AI first platform that helps smallholder farmers in developing countries. PlantVillage has an AI assistant called Nuru (Swahili for ‘light’) that has three components to its artificial intelligence: human expert-level crop-disease diagnostics using computer vision; above-human capabilities in anomaly detection and forecasting based on ground- and satellite-derived data; human language comprehension, and automated responses to questions posed by African farmers. All three of Nuru’s components are powered by recent advances made in the field of deep learning, including convolutional neural networks, recurrent neural networks and Transformer networks, respectively. Nuru is used by smallholder farmers, which has led to behavioral change and improved climate-change adaptation. It also powers knowledge delivery to 9 million farmers weekly in Kenya through a partnership with the United Nations.

Shen’s research uses deep learning, a form of AI, to investigate the interactions between hydrology and other Earth systems, such as carbon cycles. The team hopes to use this work to better characterize the ways that scarce or excess water availability impacts different parts of the natural world and society. His group received a Google AI Impact grant to develop a tool to forecast landslides using AI and also has focused on developing deep learning approaches to predict soil moisture. The tools his team have be used for predictive analytics in a variety of applications.

Sun works with multiple research groups in the College of Earth and Mineral Sciences, including Cervone’s, as well as groups led by Christelle Wauthier, associate professor of geosciences and ICDS Faculty Fellow, and Melissa Gervais, assistant professor of meteorology and atmospheric science. Sun develops and applies AI techniques for a variety from applications, from improving climate modeling to forecasting wind energy availability to improving ground movement prediction related to volcanic activity.

Published at Tue, 29 Sep 2020 18:45:00 +0000

New digital engineering technologies said to deliver up to 30% cost savings

Digital-engineering and -manufacturing processes have already revolutionized how aerospace products are designed and built.

But aerospace companies are now working to bring digital technologies to higher levels, making use of new blended digital tools capable of generating double-digit savings, say experts who spoke on a FlightGlobal-hosted digital-engineering webinar on 29 September.

But challenges exist, including the difficulty of convincing business managers to invest in new systems amid the remarkably severe aerospace industry downturn.

“Our aim is to halve the time it takes to get our products to market,” BAE Systems manufacturing technology director Andy Schofield said during the webinar. Technology enables the “taking of significant lead times out of manufacturing details and processes”.

Some customers of aerospace technology company Siemens Digital Industries Software are “consistently” achieving 20-30% reductions in development cycle times, says Siemens vice-president of aerospace and defence Dale Tutt.

Savings can compile with each generation of new products and technologies. “It really starts to add up,” he says.

Defense companies and major civil aircraft makers have been using digital system for decades. Boeing, for instance, brought such technologies to bear on development of its 777 several decades ago.

Initially, digital systems enabled transition from physical design “drafting boards” to computer-aided designs, and also involved digitising formerly-paper records.

New systems employ machine learning, artificial intelligence and digital-twin capabilities.

Critically, aerospace companies are increasingly blending various technologies together, enabling different design and development teams to access various data. Companies are also bringing suppliers onboard – giving them access to systems, which benefits the broader manufacturing ecosystem.

Manufactures can increasingly sidestep the process of “extracting” relevant data and sending that data to suppliers. Rather, technology allows suppliers to “reach in and grab” the data they need, says Paul Niewald, Boeing chief T-7A engineer and senior director.

Boeing brought “product-definition software” to bear in developing its T-7A trainer. Nodding to technology used to create the T-7A’s, the US Air Force recently bestowed the letter “e” to the aircraft’s name, calling it the eT-7A.

The move is part of a broader USAF effort to add an “e” to programmes that make full use of digital-engineering technology.

“Digital engineering allowed us to… connect a lot of things together that previously we hadn’t, so that we all had that information at the same time and were all working with the most-current data set,” Niewald says. “We kept in cadence so that one team didn’t get out in front of the other… You always had this loop back.”

“The thing that has really transformed the way we are designing and building airplanes is moving individual elements to… comprehensive digital plans… connected by digital twins,” adds Siemens’ Tutt.

Experts also expect design, development and manufacturing technologies will become “more virtual” in the coming decades, and will include more artificial intelligence.

“Designers will get to see the impact of their design downstream,” Niewald says. “We are going to see more compression of times… We are going to see these programmes continue to get faster.”

BAE’s Schofield expects companies will shift to “autonomous and automated movement of parts through the factory”.

But advancing to higher degrees of digitalization can require significant investment, and fighting for funds during a downturn is no easy task.

“Gone are the days where’s there’s a lot of discretionary funds to do R&D,” says Niewald. “The business case needs to be put together… It is difficult.”

Published at Tue, 29 Sep 2020 18:33:45 +0000