Argonne scientists promote FAIR standards for managing artificial intelligence models

Newswise – A new data standard has been developed for the AI ​​model.

Aspiring bakers are often called upon to adapt award-winning recipes based on different kitchen layouts. For example, anyone could use an eggbeater instead of a stand mixer to make an award-winning chocolate cookie.

Being able to reproduce recipes in different situations and with different preparations is important for both talented chefs and computer scientists who later face similar problems of adaptation and production. Repeat their “formula” when trying to verify and work with. New AI model. These models range from scientific applications to climate analysis to brain research.

“When we talk about data, we have a real understanding of the digital assets we deal with,” said Eliu Huerta, a scientist and director of Translational AI at Argonne National Laboratory of the U.S. Department of Energy (DOE). “With the AI ​​model, it’s a little less clear. Are we talking about data designed in a smart way, or is it a computer or software or a mix? ?

In a new study, Huerta and his colleagues demonstrated a new standard for AI model management. Adapted from recent research on automated data management, these standards are called FAIR, which stands for Accessible, Interoperable, and Reusable Search.

“By making the AI ​​model fair, we do not have to build each system from the ground up all the time,” said Argonne computer scientist Ben Blaiszik. “It’s easier to reuse ideas from different groups, helping to create cross-pollination.”

According to Huerta, the fact that many AI models today are not fair poses a challenge to scientific discovery. “For the many studies that have been done so far, it is difficult to get and Reproduce the AI ​​model referenced in the literature. “By creating and sharing FAIR AI models, we can reduce the number of duplicates of effort and share best practices for how to use these models to enable great science.”

Also Read :  Foreign Trade – The Presidential Prayer Team

To meet the needs of a diverse user community, Huerta and his colleagues have combined a unique suite of data management and high-performance computer platforms to create FAIR protocols and quantify “FAIR-ness” of AI models. . Researchers matched FAIR data published in an online repository called Materials Data Facility with a FAIR AI model published in another online repository called Data and Learning Hub for Science, as well as with AI and computer resources. Modern Argonne Leadership Computing Facility (ALCF) In this way, researchers can create computational frameworks that help connect different hardware and software, create AI models that can work the same across platforms, and it will produce results. Reusable. ALCF is the DOE Office of Science.

The two keys to creating this framework are platforms called funcX and Globus, which allow researchers to access high-performance computing resources directly from their laptops. “FuncX and Globus can help bridge the gap in hardware architecture,” said co-author Ian Foster, director of data science and studies at Argonne. “If someone is using one computer architecture and someone else is using another, now we have a simple way of speaking AI. “It’s a big part of making AI more interactive.”

In the study, the researchers used an example data set of AI models that diverted data from Argonne’s Advanced Photon source and also used the DOE science office. To perform the calculations, the team used the SambaNova system of ALCF AI Testbed and NVIDIA GPUs of Theta supercomputer (graphics processor).

Marc Hamilton, NVIDIA Vice President for Architecture and Solutions Engineering, said: High-performance computers. ” “Together, we support the global expansion of high-performance computing, which combines experimental data and edge device operations with AI to accelerate scientific discovery.”

Also Read :  Ask yourself - could you really live without your mobile phone? I had to!

Jennifer Glore, Vice President of Client Engineering at SambaNova Systems, added: “SambaNova is excited to partner with researchers at Argonne National Laboratory to continue innovating in the interface of AI and the evolving hardware architecture.” “AI will play an important role in the future of scientific computing, and the development of fair principles for AI models along with novel tools will provide Power to researchers to enable large-scale autonomous discovery. We look forward to continuing our collaboration and development at the ALCF AI Testbed. ”

A document based on the study “Principles of Justice for AI Models with Practice for Acceleration of High-Power Conversion Microscopes” appeared in Natural Science Data on 10 November 2022.

In addition to Huerta, other authors of the study include Nikon Ravi of Argonne, Pranshu Chaturvedi, Zhengchun Liu, Ryan Chard, Aristana Scourtas, KJ Schmidt, Kyle Chard, Ben Blaiszik and Ian Foster.

The research is funded by the DOE Advanced Scientific Research Office, the National Institute of Standards and Technology, the National Science Foundation and the Research and Development Assistance, led by the laboratory.

Argonne Leadership Calculator Provide advanced computing capabilities to the scientific and engineering community to promote basic discovery and understanding in a wide range of disciplines. Supported by the US Department of Energy (DOE) Office of Advanced Computer Science Research (ASCR), the ALCF is one of two DOE leadership calculators in the country dedicated to open science.

About Advanced Photon Source

Advanced Photon (APS) of the US Department of Energy’s Office of Energy Science at the Argonne National Laboratory is one of the world’s most productive X-ray sources. APS provides high-light X-rays to a diverse community of researchers in the sciences, materials, chemistry, physics, condensed matter, life and environmental sciences, and applied research. These X-rays are ideally suited for the exploration of materials and biological structures. Item distribution; Chemical, magnetic, electronic states; And many technologically important engineering systems, from batteries to sprayers, which are fundamental to our country’s economic, technological and physical well-being. Each year, more than 5,000 researchers use APS to produce more than 2,000 publications detailing the findings, influencing and addressing key biological protein structures more than users of other X-ray light sources. APS scientists and engineers develop new technologies that are central to the process of accelerating and operating light sources. This includes high-intensity X-ray input devices awarded by researchers that focus X-rays down a few nanometers, devices that enhance the way X-rays interact with the model being studied, and applications that gather and Manage large volumes of data obtained from research findings at APS.

Also Read :  Kakao's co-CEO steps down amid public anger over chat app outage

This research utilizes the resources of Advanced Photon, a U.S. Office of Scientific Consumer DOE, operated for the DOE Scientific Office by the Argonne National Laboratory under contract number DE-AC02-06CH11357.

Argonne National Laboratory Find solutions to national problems in science and technology. Argonne’s first national laboratory conducts basic research and practice in almost every science discipline. Argonne researchers work closely with researchers from hundreds of companies, universities, and federal, state, and municipal agencies to help them address their specific issues, advance American scientific leadership, and prepare the nation for a better future. With staff from over 60 countries, Argonne is managed by UChicago Argonne, LLC for the United States Department of Energy Science Office.

United States Department of Energy Science Office Is the single largest advocate of basic research in physical science in the United States and is working to address some of the key challenges of our time. For more information, visit https: // energy .gov / science.


Leave a Reply

Your email address will not be published.