Engineering | News releases | Research | Science | Technology

July 29, 2021

The UW will lead a new artificial intelligence research institute that will focus on fundamental AI and machine learning theory, algorithms and applications for real-time learning and control of complex dynamic systems, which describe chaotic situations where conditions are constantly shifting and hard to predict.Andy Freeberg/University of Washington

The U.S. National Science Foundation today announced 11 new artificial-intelligence research institutes, including one led by the University of Washington. These institutes are part of a $220 million investment spanning seven research areas in AI. Each institute will receive about $20 million over five years.

J. Nathan KutzUniversity of Washington

The UW-led AI Institute for Dynamic Systems will focus on fundamental AI and machine learning theory, algorithms and applications for real-time learning and control of complex dynamic systems, which describe chaotic situations where conditions are constantly shifting and hard to predict.

The engineering sciences are undergoing a revolution that is aided by machine learning and AI algorithms, said institute director J. Nathan Kutz, a UW professor of applied mathematics. This institute brings together a world-class team of engineers, scientists and mathematicians who aim to integrate fundamental developments in AI with applications in critical and emerging technological applications.

Researchers know the basic physics principles behind dynamic systems, which include situations such as turbulence or how the body recovers from an injury. But these scenarios are often happening on multiple timescales at once and can be a combination of many types of physics, making it hard for researchers to understand exactly whats going on.

My favorite dynamic system is turbulence, said institute associate director Steve Brunton, a UW associate professor of mechanical engineering. We literally live and breathe inside of a working fluid, and so do nearly all of our machines. But because of the multiscale complexity of the fluid, which involves a cascade of increasingly smaller eddies, we still have an incredibly hard time predicting what fluids will do outside of idealized and controlled settings.

The overall goal of this institute is to integrate physics-based models with AI and machine learning approaches to develop data-enabled efficient and explainable solutions for challenges across science and engineering.

Some of our specific questions include: Can we develop better machine-learning technologies by baking in and enforcing known physics, such as conservation laws, symmetries, etc.? Brunton said. Similarly, in complex systems where we only have partially known or unknown physics such as neuroscience or epidemiology can we use machine learning to learn the physics of these systems?

In addition to research, the institute will be focused on training future researchers in this field throughout the education pipeline. Some examples include: partnering with high school programs that focus on AI-related projects and creating a post-baccalaureate program that will actively recruit and support recent college graduates from underrepresented groups, United States veterans and first-generation college students with the goal of helping them attend graduate school.

The institute will provide massive open-source educational materials that include lectures, data and code packages for advancing and empowering AI, Kutz said. Importantly, we will provide AI ethics training for all involved in the institute. We will also make this training available to the community at large, thus enforcing a disciplined approach to thinking about AI and its implications for our emerging societal concerns around data, data privacy and the ethical application of AI algorithms.

As part of the educational component, the team will use a lightboard (Steve Brunton shown here) to create a range of high-quality educational and research videos focusing on key aspects of AI and machine learning for engineering dynamical systems and control. Educational content will be made freely available to the community on YouTube.Dennis Wise/University of Washington

For this institute, the UW is partnering with several regional institutions the University of Hawaii at Mnoa, Montana State University, the University of Nevada Reno, Boise State University, the University of Alaska Anchorage and Portland State University as well as with Harvard University and Columbia University.

We are so excited to bring together a critical mass of amazing and innovative researchers from across the U.S. to really move the needle in developing machine learning technology for physical and engineering dynamic systems, Brunton said. We also have a deep connection with industry partners, such as Boeing, which provides us with an incredible opportunity to make sure that we are focused on important and relevant problems and that our technology will actually be used.

Additional UW researchers who are part of this institute are lead researcher Krithika Manohar, assistant professor of mechanical engineering; Maryam Fazel, professor of electrical and computer engineering; Daniela Witten, professor of biostatistics; and David Beck, a research associate professor of chemical engineering.

Im glad to see this substantial investment going to one of our states premier research institutions, said U.S. Sen. Patty Murray, D-Wash. As the University of Washington and other research institutions in our state continue to lead on artificial intelligence, this investment will be critical to ensuring that the state of Washington remains a leader in innovation, research and scientific achievement. Ill keep fighting for important federal investments like this one to move this work forward.

The UW is also a partner institution on another newly announced NSF institute, the AI EDGE Institute, which is led by Ohio State University. The goal of this institute is to design future generations of wireless edge networks that are highly efficient, reliable, robust and secure.

These 11 new AI institutes are building on the first round of seven AI institutes funded in 2020, and expand the reach of these institutes to include a total of 40 states and Washington D.C.In addition to the UW-led institute, the state of Washington will also house the Institute for Agricultural AI for Transforming Workforce and Decision Support, or AgAID Institute, led by Washington State University. The institutes goal is to use AI to tackle some of agricultures biggest challenges related to labor, water, weather and climate change.

The state of Washington is already a leader in artificial intelligence, said U.S. Sen. Maria Cantwell, D-Wash. From the University of Washingtons Tech Policy Lab that studies the grand challenges around artificial intelligence to Washington State Universitys work in precision agriculture, we are more than ready for these two grants to help us understand more artificial intelligence applications.The UW will work in the area of complex systems to improve fields like manufacturing, and WSU will work on improvements in farming.

The AI Institute for Dynamic Systems is partially funded by the U.S. Department of Homeland Security.

For more information, contact Kutz at kutz@uw.edu and Brunton at sbrunton@uw.edu.

See original here:
UW to lead new NSF institute for using artificial intelligence to understand dynamic systems - UW News

While the country continues to struggle convincing people it is safe and smart to get vaccinated against Covid-19 despite emergency authorization and advocacy from the U.S. Federal Drug Administration (FDA), other growing numbers of people are willing to jump in with both feet when it comes to trusting their lives to Artificial Intelligence (AI), which has no standards or organizational oversight. At this time, only guidelines exist from the U.S. Federal Trade Commission (FTC) for AI.

I will take you through an example of how even impactful applications of AI still need a human assistant to ensure trustworthy, explainable, and unbiased decision making.

Internet law concept with 3d rendering ai robot with law scale

Using AI for automating manufacturing and improving our work streams, such as providing robust CMMS (computerized maintenance management systems) is an excellent application for AI. CMMS is a proactive methodology to keep systems running and to optimize maintenance operations. Alternatively, if businesses and institutions dont invest in smart maintenance approaches and choose only to take action after a system fails, it ultimately impacts operations, and customer service. Especially, if the components needed for repair are not readily available.

CMMS systems utilize sensors and other Internet of Things (IoT) devices and leverage on cloud services. They can track equipment assets, sense whether device performance is degrading, maintain the history of repair times and help us determine whether it is time to invest in replacing or upgrading systems versus investing in more repairs. AI can be used to look at this plethora of data and help make inferences that could save money, anticipate risk factors and build risk mitigation plans.

So far in this example, it appears that an individual human cant be adversely affected by the AI used in this CMMS application. However, lets consider a scenario when a relationship appears between the number of times a repair was done on a device that repeatedly failed and the person who performed that repair.

Does this indicate that the repair person needs more training or does it simply mean the device has too many faults and should be replaced? This is when the AI needs to ask a human for help and the human needs to be smart about interpreting what the data means.

Whenever AI makes a decision on a humans livelihood or is used to evaluate an individuals competency, we need to hit the big red STOP button and understand how the AI is making decisions and what it means. There have been many examples of companies using AI to evaluate employee job performance, which resulted in the subsequent dismissal or poor evaluations of employees. Lending institutions using AI have encountered algorithms that have unfairly eliminated underrepresented groups of individuals from qualifying for loans.

This kind of blind faith in computer generated decisions gives me a flashback to high school when they gave us career assessment tests to match us to a future vocation.

I was at the top of my class and my best friend, a male had scored slightly lower than I did in math and science. We were both expecting the outputs of our career assessments to be similar. How wrong I was!

His assessment said he could become an engineer, scientist, or politician. The results of my assessment, said I could be a cook, or sell cosmetics.

True fact:I stink at cooking. I never could cook and I still cant. Furthermore, it would be cruel and unjust for anyone to be subjected to eating my cooking.

When I complained to the guidance counselor, he reminded me that the results had to be correct, because this was a computer generated result and therefore it had to be accurate. Thankfully, I have always been a rebel and I ignored that assessment and went on to become an engineer.

AI has so much potential, but it has a long way to go before it can be considered a standalone replacement for human decision making that is trustworthy and unbiased.

In a recent IEEE/IEEE-HKN webinar, Dr.Manuela M. Veloso, Head of J.P. Morgan AI Research discussed human assisted AI as the bridge in the quest to help create more robust trustworthy AI. Including the human in the loop while the AI is exploring the data and asking the human for help is a new paradigm that many have leaped over to expedite the use of AI. Its time to embrace that Human Assisted AI is necessary if we are to move forward developing robust applications of AI. We need to demystify AI as a building block and have a way for the AI to declare it needs help from its human partner.

Finally, we need to be vigilant questioning the decisions from AI and help lawmakers develop standards to prevent blind trust in computer generated outputs that could have disastrous impacts of turning aspiring engineers into horrible cooks.

The rest is here:
Human Assisted Artificial Intelligence: A Pathway to Trustworthy And Unbiased AI - Forbes

Humans have been seeking the answer to a colossal challenge for half a century: identifying the basic building blocks of life, essential knowledge in the battle against terminal illnesses. Water is easy to imagine. It is simply two hydrogen atoms connected to another of oxygen: HO. But the protein that enriches blood, hemoglobin, carries the less memorable formula CHNOSFe. In 1969, US biologist Cyrus Levinthal calculated that it would require more time than has passed since the formation of the universe around 14 billion years to untangle all the potential configurations of a single protein from its amino acid sequence, which are the links of these macromolecules. However, an artificial intelligence system created by the Google conglomerate has achieved this feat in just a few minutes. Its predictions for practically every single human protein were published last week in a giant step for biology that has removed a blindfold from human knowledge.

At the vanguard of this revolution is British neuroscientist Demis Hassabis. The 44-year-old researcher was a chess child prodigy who in 1997 was left in awe of the match between Russian grandmaster Garry Kasparov and the super computer Deep Blue. The machine beat the man, but Hassabis was left with the impression that it was a cumbersome piece of junk, and useless in a game of tic-tac-toe. When the final game was over, the University of Cambridge undergraduate came up with the idea of designing a machine that could learn any game. In 2010, Hassabis founded a company called DeepMind to lead the investigative drive toward Artificial Intelligence (AI). In 2013, his first creation taught itself to play and win at various games on the legendary Atari console. In 2014, Google bought DeepMind for $650 million (around 500 million at the exchange rate of the time).

After limbering up on video games, the scientists at DeepMind then set themselves the task of solving one of the greatest challenges in biology. Proteins like hormones, enzymes and antibodies are tiny machines that carry out the basic functions of life. They are made up of chains of smaller molecules, amino acids, much like a pearl necklace. These necklaces are folded in convoluted configurations that determine their function. Antibodies, the human bodys defense mechanism against invaders like the coronavirus, have a Y shape.

The recipes of all the proteins required to function are written in the DNA of every cell. The DeepMind system, baptized AlphaFold, reads this information a sequence of amino acids and predicts the structure of each protein. Its precision is similar to that achieved in laboratory experiments, which require considerably more time and money. It is like guessing the structure of a quiche after seeing pie crust, eggs, pepper, salt, milk and cream for the first time.

DeepMind and the European Molecular Biology Laboratory (EMBL) published more than 350,000 structures on July 22, including some 20,000 human proteins and those of 20 other organisms, such as a lab mouse and the tuberculosis bacteria. Venki Ramakrishnan of the Medical Research Council Laboratory of Molecular Biology in Cambridge and winner of the 2009 Nobel Prize in Chemistry, says that is an astonishing advancement with unpredictable consequences. It has taken place long before many experts had predicted. It is going to be exciting to see the many ways in which it is going to radically change biological investigation.

Some organizations are already working with the new database. The Drugs for Neglected Diseases initiative, a global non-profit set up with the aid of Doctors Without Borders, uses the structure of proteins to seek new treatments. Practically all diseases, from cancer to Alzheimers, and including Covid-19, are related to the structure of one protein or another. Other institutions, such as the University of Portsmouth in the UK, are using the program to try and design proteins capable of recycling plastic.

Hassabis, executive director of DeepMind, has announced the companys plan is to publish 100 million structures over the next few months. The idea is to offer the predictions for the structure of practically every protein with a known sequence of amino acids free of charge. We believe that this is the most important contribution to date that artificial intelligence has contributed to scientific knowledge, he said following the publication of DeepMinds research in the medical journal Nature.

The AlphaFold system was not created from nothing, as Edith Heard, director general of the EMBL, has emphasized. AlphaFold has been trained using data from public resources developed by the scientific community, so it makes sense that its predictions are also made public. In Heards view, the system represents a genuine revolution in life sciences, like the genome was decades ago.

To determine the real structure of a protein, hugely expensive infrastructure is required, such as the European Synchrotron Radiation Facility, a particle accelerator covering almost a square kilometer in Grenoble, France. The radiation emitted by the electrons that circle the ring, which basically consist of X-rays, allow researchers to observe the secrets of matter. Spanish biologist Jos Antonio Mrquez explains that elucidating the shape of a protein with a synchrotron, or with the alternative method of cryogenic electron microscopy, could require months or even years. AlphaFold can achieve it in minutes, albeit with errors.

We are talking about computer-generated predictions, not the experimental determination of the structure. And the precision is 58%, says Mrquez, a 52-year-old researcher who heads the Crystallography Platform at the EMBL in Grenoble. As things stand, if a scientist wants to study a protein connected to cancer, it could take months or years to analyze its structure. There are only around 180,000 structures in available databases. The information published by DeepMind has doubled that number. And in a few months millions will be available. Today it is common to not find a protein in the databases. With AlphaFold you can obtain a prediction with a 58% reliability. You save an enormous amount of time, says Mrquez, who did not participate in the project. The systems imprecisions are concentrated in specific regions of the proteins, which are unstructured to adapt to the environment.

Mrquez points out other limitations. The DeepMind system can predict the structure of an isolated molecule, but proteins tend to interact with other proteins. AlphaFold is not yet capable of predicting the structure of these complexes but it is a system designed to learn on its own. Mrquez is optimistic: It will speed up discoveries in practically all areas of biology.

Read this article:
Artificial intelligence uncovers the building blocks of life and paves the way for a new era in science - EL PAS in English