The Department of Energy’s Oak Ridge National Laboratory hosted its annual Smoky Mountains Computational Sciences and Engineering Conference in person for the first time since the COVID-19 pandemic.
SMC23 took place in downtown Knoxville, TN from August 29–31 and boasted nearly 200 attendees. The first gathering took place 20 years ago at Fall Creek Falls State Park and hosted a fraction of the conference-goers.
This year’s theme: “Peering into the future: Accelerating science and engineering discoveries through advanced technologies, integrated research infrastructures for experiments, machine learning/AI, and modeling and simulation.”
The AI focus was particularly important as for the second year in a row the conference kicked off with an AI workshop. This year’s “Generative AI for ORNL Science Workshop” featured nearly 100 attendees and provided a venue to educate and exchange research and development ideas, collaborations, and investments around the current state-of-the-art in these relatively new technologies with enormous potential to transform society.
“One of the goals of the AI workshop and our SMC23 theme is to understand our next-generation applications and opportunities for scientific discovery that will leverage AI, modeling and simulation, and interconnected instruments,” said conference organizer and ORNL Senior Computer Scientist Oscar Hernandez. “The new frontier is where AI will accelerate our next-generation applications harnessing exascale computing, but we need to make sure that the use of AI is safe, trustworthy, and efficient, as many of our applications are mission-critical,” he added.
Following the workshop, attendees enjoyed dinner and a keynote talk by the University of Montreal/Mila-Quebec AI Institute’s Irina Rish, who discussed “The Large-scale AI Revolution: Breakthroughs and Challenges.” Rish laid out current trends in AI, such as the generative AI boom that has brought us ChatGPT and AI image generators such as DALL-E, and their role in the achievement of the “Holy Grail” of generalized AI, which OpenAI defines as “highly autonomous systems that outperform humans at most economically valuable work.”
To get there, Rish says, we need to overcome numerous challenges including scaling, cost-effectiveness, and training, all of which will require broad collaborations.
Two days of talks, networking, and collaboration followed, with many attendees duking it out on the bowling lanes of Maple Hall in downtown Knoxville to close out the second night.
“Our goal was to recreate the informal setting we’ve historically enjoyed at SMC as we discuss the state-of-the-art and upcoming scientific priorities and challenges,” conference co-chair Arjun Shankar said.
The conference also featured its popular “Data Challenge,” in which six finalists emerged from a total of 35 registrants.
Contestants could choose among seven challenges covering a range of topics, including “Diamonds and ice: when is H2O no longer a molecule?,” “Sustainable Cities: Socioeconomics, Building Types, and Urban Morphology,” and “Automating signal extraction – scientist vs computer.”
“Our student teams really appreciated the chance to work on real world data again this year and we had a diverse mix of competitors from High School Student to Graduate Students,” said Hong Liu, SMC Data Challenge Chair and an ORNL HPC engineer.
C for Comp Sci from nearby Farragut High School won the Best Paper Novice category by completely solving Challenge 1, “Machine Learning approaches to High Throughput Phenotyping.” A team featuring Yasmeen Haleem from the University of North Texas and Isabelle Wagenvoord from Colorado College won Best Poster by solving Challenge 7, “EAGLE-I Outage Data 2014-2022.” Other winners included Best Lighting Talk by Ruhaan Singh, also of Farragut High School, for his solution to Challenge 7.
“It was great to work with the data sponsors to provide this opportunity for the seventh year of the SMC Data Challenge,” said ORNL HPC Engineer and SMC Data Challenge Co-Chair Suzanne Parete-Koon. “Activities like this help make students and researchers aware of the National Labs and the science that is done.”
This year represented the first time the conference was held in downtown Knoxville, and attendees enjoyed the energetic urban setting.
“SMC is a testament to ORNL’s leadership in driving the future of computing and data,” said ORNL Interim Associate Lab Director for Computing and Computational Sciences Shaun Gleason. “It was a great experience to have everyone back in person, and we look forward to an even bigger, better event next year."
SMC23 was sponsored by Intel, Samsung, SambaNova Systems, Hewlett Packard Enterprise, Groq, AMD, NVIDIA, Penguin Computing, IBM, Cornelis Networks, and CTG Federal.
UT-Battelle manages ORNL for the Department of Energy’s Office of Science, the single largest supporter of basic research in the physical sciences in the United States. The Office of Science is working to address some of the most pressing challenges of our time. For more information, please visit energy.gov/science.