The four new AAAS fellows among the 91探花faculty are:

, professor emeritus in the Paul G. Allen School of Computer Science & Engineering, is honored for contributions to artificial intelligence and machine learning. Domingos is particularly known for his introduction of Markov logic networks, which presented a simple yet efficient approach to unifying first-order logic and probabilistic reasoning to support inference learning. He also helped pioneer the field of adversarial learning, producing the first algorithm to automate the process of adversarial classification to enable data mining systems to adapt rapidly against evolving adversarial attacks. Domingos subsequently contributed the first unsupervised approach to semantic parsing, which enables machines to extract knowledge from text and speech, a process that underpins machine learning and natural language processing. In 2015, he published 鈥�,鈥� a book that examines how machine learning increasingly influences every aspect of people鈥檚 lives. Domingos joined the 91探花faculty in 1999 and remains active in research after attaining emeritus status earlier this year.

, professor in the Department of Physiology and Biophysics, is a pioneer in brain-machine interfaces. His earlier work was on the brain鈥檚 direction of arm and leg movements. Fetz later showed that the brain could volitionally control certain nerve cells, called cortical neurons, in various patterns. This became the foundation for research on the unexpected ability of neural activity to drive external devices. Fetz also conducted studies of interneurons in the spine, and demonstrated that they had many properties of cells in the cortex, including their preparation to carry out instructed movements. Fetz also developed dynamic network models to simulate neural interactions that target tracking and short-term memory. In an historical achievement, his lab designed and tested an implantable neurochip that can record activity of cortical cells and convert this in real-time to stimulate the cortex, spinal cord or muscles. The brain can learn to incorporate this artificial feedback loop into behaviors. The neurochip holds future promise for clinical applications, such as moving paralyzed muscles.

is a professor in the Department of Anesthesiology and Pain Medicine, as well as a professor in the Public Health Sciences Division at the Fred Hutchinson Cancer Research Center. Raftery studies the small molecules at work during metabolism in cells, animals and people. He has developed analytical and statistical methods to profile metabolites in complex biological samples. Metabolites are the end products of many biochemical functions in living systems. Raftery鈥檚 research is working to discover sensitive biomarkers indicating the presence of disease and its progression. He has applied his advances in metabolomics to detect very early stages of cancer, as well as in his research on diabetes and heart disease. He is a scientist at the 91探花Mitochondrial and Metabolism Center, which, among its goals, is investigating the roles of cell metabolism dysfunction in common diseases and is also seeking related diagnostic and therapeutic tools. Raftery also directs the interdisciplinary Northwest Metabolomics Research Center, which fosters collaborations among scientists from several institutions. The lab uses some of the latest technologies and capabilities to improve the metabolic understanding of a variety of serious disorders.

, a professor in the Paul G. Allen School of Computer Science & Engineering, was honored for his contributions to artificial intelligence spanning automated planning, software agents, crowdsourcing and internet information extraction, as well as his efforts to commercialize AI technologies. Weld leads the UW鈥檚 , where he focuses on advancing explainable AI to allow people to better understand and control AI-powered tools, assistants and systems and combine human and machine intelligence to accomplish more together than alone. Weld has co-founded multiple startup companies, including Netbot, Inc., which produced the first online comparison shopping engine that was subsequently acquired by Excite, and AdRelevance, an early provider of tools for monitoring online advertising data acquired by Nielsen Netratings. A member of the 91探花faculty since 1988, Weld is a venture partner and member of the Technology Advisory Board of Madrona Venture Group and Allen Institute for Artificial Intelligence, where he also leads the focused on the development of AI-powered tools to help scientists extract useful knowledge from scholarly literature.

In addition, , a professor in the Vaccine and Infectious Disease Division of the Fred Hutchinson Cancer Research Center, was selected 鈥渇or distinguished contributions to the field of HIV prevention research, particularly for design and analysis of clinical trials of pre-exposure prophylaxis and treatment as prevention.鈥� Donnell is also a 91探花affiliate of global health and of health services.

Researchers at the 91探花 share their expertise on artificial intelligence and data science in “,” an online documentary produced and released this winter by YouTube. The series narrated by Robert Downey Jr. looks at how AI could affect everything from health care to the search for extraterrestrial life.

, professor in the Paul G. Allen School of Computer Science & Engineering, is a recurring expert who offers commentary in several episodes. In 2015 Domingos published “,” a popular book about the promise of artificial intelligence.

The seventh episode, titled “,” features the UW-based . After looking at elephant poaching and new plant-based foods, the segment looks at how seismologists are collecting and processing data to warn of incoming earthquakes along the Cascadia subduction zone. (Domingos first appears in the episode , and the earthquake segment begins .)

, director of the PNSN and a professor of Earth and space sciences, strolls through downtown Seattle and discusses the challenges and prospects for long-term earthquake prediction. , research professor of Earth and space sciences, describes how the 91探花system identifies shaking generated by seismic events, and , a field engineer and lab coordinator with the PNSN, shows off a seismic monitoring station near the Space Needle.

The series can be streamed free with advertisements, or ad-free for YouTube subscribers. Earlier episodes have been viewed millions of times.

A novelist’s thoughts on storytelling, a geologist’s soil restoration strategy, an environmentalist’s memoir, a celebration of Latino music influences, a poet’s meditations on her changing city 鈥�

Yes, and a best-selling author’s latest work, a podcast reborn as a book, a collaboration of world-class violists and even tales of brave Icelandic seawomen 鈥� at this festive time of year, 91探花 faculty creations can make great gifts for the Dawg on your shopping list.

Here鈥檚 a quick look at some gift-worthy books and music created by 91探花talents in the last year or so 鈥� and a reminder of some perennial favorites.

Charles Johnson, “.” Johnson, National Book Award-winning author of “” and longtime professor of English, discusses his art in a book stemming from a year of interviews. “There is winning sanity here,” the New York Times wrote: “Johnson wants his students to be ‘raconteurs always ready to tell an engaging tale,’ not self-preoccupied neurotics.” Published by .

Marisol Berr铆os-Miranda, Shannon Dudley and Michelle Habell-Pall谩n, ““ An extraordinary exhibit at the Smithsonian and Seattle’s Experience Music Project (now Museum of Pop Culture) comes to life as a book, detailing Latino influence on American popular music from salsa to punk, Chicano rock to the Miami sound. Berrios-Miranda is an affiliate associate professor of ethnomusicology, Dudley an associate professor of music and Habell-Pall谩n an associate professor in the Department of Gender, Women and Sexuality Studies. It’s a dual-language volume 鈥� English on the right side, Spanish on the left. And as a bonus the authors have created an American Sabor on iTunes and Spotify; the book flags specific songs with a playlist icon. Published by 91探花 Press.

David R. Montgomery, “.” Montgomery, a professor of Earth and space sciences, won praise for his popular 2007 book “.” Several books later he returned in 2017 with this view of environmental restoration based on three ideas 鈥� “ditch the plow, cover up, grow diversity.” said Montgomery’s well-expressed views “will convince readers that soil health should not remain an under-the-radar issue and that we all benefit from embracing a new philosophy of farming.” Published by .

Margaret Willson, Willson is an affiliate associate professor of anthropology and the Canadian Studies Arctic Program. In her years working as a deckhand she came across historic accounts of a woman sea captain known for reading the weather, hauling in large catches and never losing a crew member in 60 years of fishing. “And yet people in Iceland told me there had been few seawomen in their past, and few in their present,” she said. “I found this strange in a country of such purported gender equality. This curiosity led to the research and all that came from it.” Published by .

Estella Leopold, “.” Leopold is professor emeritus of biology and the youngest daughter of , who wrote the 1949 classic of early environmentalism, “.” She returns to scenes of her Wisconsin childhood in this follow-up, describing her life on the land where her father practiced his revolutionary conservation philosophy. Published by .

David Shields, “.” Shields is a professor of English and the best-selling author of many books, starting with his 1984 novel “.” In 2017 he brought out this collection of essays that the New York Times called “a triumphantly humane book” and him “our elusive, humorous ironist, something like a 21^st century Socrates.” The paper’s praise continued: “He is a master stylist 鈥� and has been for a long time, on the evidence of these pieces from throughout his career. . . All good writers make us feel less alone. But Shields makes us feel better.” Published by .

Joseph Janes, “.” The year 2017 saw Janes’ popular podcast “” become a book under a slightly different title. Janes is an associate professor in the Information School who writes here about the origin and often evolving meaning of historical documents, both famous and less known. Some of his favorite “documents” are Sen. Joseph McCarthy’s fictional list of communists, the Fannie Farmer Cookbook and the backstory to what’s called the Rosie the Riveter poster. Published by .

Frances McCue, Well-known Seattle poet, teacher and self-described “arts instigator,” McCue is a senior lecturer in English. She was a co-founder of Hugo House, a place for writers, and served as its director for 10 years. Those experiences fuel this book of poems about the changing nature of the city. “This is Seattle. A place to love whatever’s left,” she writes. Published by .

Scott L. Montgomery, “.” Scientific research that doesn鈥檛 get communicated effectively to the public may as well not have happened at all, says geoscientist Montgomery in this second volume of a popular 2001 book. A prolific writer, Montgomery is a lecturer in the Jackson School of International Studies. “Communicating is the doing of science,” he adds. “Publication and public speaking are how scientific work gains a presence, a shared reality in the world.鈥澛� Published by .

Odai Johnson, “.” The true cultural tipping point in the run-up to the American Revolution, writes Johnson, a professor in the School of Drama, might not have been the Boston Tea Party or even the First Continental Congress. Rather, he suggests, it was Congress’ 1774 decision to close the British American theaters 鈥� a small act but “a hard shot across the bow of British culture.” Published by .

Here are some recordings from 2017 involving faculty in the 91探花School of Music:

Melia Watras, “.” Music professor Watras offers a collaboration from of world-class violists performing and sharing their own compositions with each other. Her own playing has been described in the press as “staggeringly virtuosic.” Richard Karpen, School of Music director, is among several guests. The title comes from the number of strings on the instruments used: two violas, one violin, and the 14-string viola d’amore. .

Cuong Vu 4-Tet, “.” A live collaboration between Vu, 91探花Jazz Studies chair, and renowned jazz guitarist Bill Frisell, who is an affiliate professor with the School of Music. Recorded in 2016 at Meany Theater, Vu and Frisell were joined by artists in residence Ted Poor on drums and Luke Bergman on bass. Released on .

Craig Sheppard, “.” Sheppard, longtime professor of music and a world-class pianist, plays sonatas and mazurkas by Frederic Chopin recorded live at Meany Theater in February 2017. The Seattle Times said of an earlier Chopin concert of Sheppard’s that his playing featured “exquisite details 鈥� it was playing that revealed layer after layer of music in each piece, as if one were faceting a gemstone. Released on .

Here are some other notable recent UW-authored books:

• Research on poverty and the American suburbs in “,” by Scott Allard, professor in the Evan School of Public Policy & Governance.
• Literature meets science to contemplate the geologic epoch of humans in “,” co-edited by Jesse Oak Taylor, associate professor of English.
• A popular science exploration of machine learning and the algorithms that help run our lives in “,” by Pedro Domingos, professor of computer science and engineering.
• A close look at four of America’s electoral adventures in “” by Margaret O’Mara, professor of history.
• A fully revised second edition of Earth and space sciences professor Darrel Cowan’s popular 1984 book, “.” This 378-page paperback is filled with details about Washington state geology.
• The story of a city’s transition from the Ottoman Empire to Greece in “” by Devin Naar, professor of history and Jewish studies.
• A city that “thinks like a planet” is one both resilient to and ready for the future that the changing Earth will bring, says Marina Alberti, professor in the College of Built Environments in “.
• Todd London, professor and director of the School of Drama, follows the professional theater experiences of 15 actors from the 1995 class of Harvard’s American Repertory Theater in “.”
• Dr. Stephen Helgerson, a 91探花School of Public Health alumnus and physician in preventive medicine for four decades, uses the novella form to tell of the influenza epidemic’s arrival in his state in “.”
• On the 500^th anniversary of the Protestant Reformation, an exploration of faith that results in the common good in 鈥�,鈥� co-authored by Steve Pfaff, professor of sociology.
• Calm down from holiday 鈥� and tech-induced stresses 鈥� by thinking mindfully with “” by communication professor David Levy.

Finally, still-popular and pertinent books from a few years back include the second edition of “” by Jeffrey Ochsner, professor of architecture; “” by Randlett with Frances McCue; “” by Cliff Mass, professor of atmospheric sciences; and the ever-popular “” by Bill Holm, professor emeritus of art history. All of these were published by , which has many other great titles.

Recognizing that, the White House Office of Science and Technology Policy is co-hosting four public workshops on artificial intelligence 鈥� the first of them May 24 at the 91探花. Subsequent events will take place in ; in ; and in .

Put on by the and the 91探花, the will focus on legal and policy issues around artificial intelligence, or AI.

Speakers include:

• , law school dean and president of the Association of American Law Schools
• , special assistant to the president for economic and technology policy
• , White House deputy U.S. chief technology officer
• , a 91探花assistant professor of law and co-director of the Tech Policy Lab
• , a 91探花professor of computer science and engineering and author of 鈥溾��
• , chief executive officer of the Allen Institute for Artificial Intelligence and a 91探花professor of computer science and engineering
• , an associate professor in the School of Information at UC Berkeley and co-director of the Berkeley Center for Law & Technology
• , a principal researcher at Microsoft Research New York City and senior researcher at NYU Information Law Institute
• , a law professor at Yale Law School
• Camille Fischer, policy advisor, National Economic Council
• Terah Lyons, policy advisor, White House Office of Science and Technology Policy

Etzioni will provide an overview on the current state of artificial intelligence, followed by two panel discussions. The first will examine issues around making decisions in the private or public sector using artificial intelligence.

The second panel will focus on logistical aspects of AI applications, such as when the government might reasonably feel comfortable turning mail delivery over to robots or how safe autonomous flight must be to be used for deliveries.

The aim of the workshops is to look at the advantages and drawbacks of artificial intelligence. As a White House points out, President Obama鈥檚 and the will both rely on AI to identify patterns in medical data and help doctors diagnose diseases and determine treatment plans. But others worry the technology will displace human workers, or go so far as to that it could pose a threat to the human race.

The 91探花workshop, free and open to the public, will be held from 1:30 to 5 p.m. May 24 in the Magnuson Jackson Courtroom 138 at the 91探花School of Law. A reception follows from 5 to 7 p.m. Registration is available , and the conference will be .

The next in the series, about artificial intelligence for social good, is June 7 in Washington, D.C., followed by a June 28 on safety and control for AI at Carnegie Mellon University in Pittsburgh and a July 7 in New York City on the social and economic implications of AI.

For more information, contact Ryan Calo at rcalo@uw.edu or 206-543-1580.

A popular science romp through one of today’s hottest scientific topics, is an essential primer on machine learning. It unveils the deep ideas behind the algorithms that increasingly pick our books, find our dates, filter email, manage investments and run our lives 鈥� and what informed consumers and citizens ought to know about them.

Domingos, who will at 7:30 p.m. on Sept. 22, answered a few questions about the book.

What is machine learning, and how might a person encounter it in a typical day?

笔顿:听 Machine learning is the automation of discovery 鈥� computers learning by themselves by generalizing from data instead of having to be programmed by us. It鈥檚 like the scientific method on steroids: formulate hypotheses, test them against the data, refine them 鈥� except computers can do it millions of times faster than humans.

Google uses machine learning to decide which Web pages to show you, Amazon and Netflix to recommend books and movies, Twitter and Facebook to select posts for your feed. Siri uses learning algorithms to understand what you say and predict what you want to do. Spam filters use it as well. Retailers use it to decide which goods to stock and how to lay out their stores. If you receive a credit card offer, chances are a learning algorithm picked you. At many companies, when you apply for a job, a learning algorithm screens your resume. Online dating sites use machine learning to match their users 鈥� there are children alive today who wouldn鈥檛 have been born if not for machine learning. In other words, machine learning is involved in pretty much everything we do these days.

Why is it important for someone who isn’t a computer scientist to understand principles of machine learning?

PD: Learning algorithms make a lot of decisions on your behalf every day. As we just saw, they can determine not just what goods you buy but also whether you’ll get a job or even who your lifetime companion will be. If these algorithms are a black box to you, you have no control over where they will take you. Think of a car as an analogy: only engineers and mechanics need to understand how the engine works, but you need to know how to drive it. In the future cars will drive themselves, but you鈥檒l have to know how to drive learning algorithms 鈥� and right now you probably don鈥檛 even know where the steering wheel or the pedals are.

Your book talks about what different “tribes” in machine learning research might contribute to curing cancer, and what their approaches lack. Why focus on that question?

PD: Curing cancer is one of the most important problems in the world 鈥� perhaps the most important problem 鈥� and machine learning has a big part to play in solving it. What makes cancer hard is that it鈥檚 not one disease, but many. Every patient鈥檚 cancer is different, and it mutates as it grows, so there鈥檚 no one-size-fits-all solution. The cure for cancer is a learning program that predicts which drug to use for which cancer by looking at the tumor鈥檚 genome, the patient鈥檚 genome and medical history, etc. But none of the current approaches to machine learning is able to solve the problem all by itself, so it鈥檚 a great illustration of both what each approach brings to the table and what it鈥檚 missing.

What is the difference between the algorithms that Netflix and Amazon use to recommend products you might like? Why is it important for consumers to be aware of these differences?

PD: Like every company, Netflix and Amazon each use the algorithms that best serve their purposes. Neflix loses money on blockbusters, so its recommendation system directs you to obscure British TV shows from the 70s, which cost it virtually nothing. The whole machine learning smarts is in picking shows for you that you鈥檒l actually like even though you鈥檝e never heard of them. Amazon, on the other hand, has no particular interest in recommending rare products that only sell in small quantities. Selling larger quantities of fewer products actually simplifies its logistics. So its recommendation system is based more on just how popular each product is in connection with the products you鈥檝e bought before. The problem for you if you don鈥檛 know any of this is that you wind up doing what the companies want you to do, instead of what you want to do.

If you know 鈥� even just roughly 鈥� how the learning algorithms work, you can make them work for you by deliberately teaching them, by choosing the companies whose machine learning agrees best with you and by demanding that the learning algorithms let you explicitly say things like “This is what I want, not that,” and “Here鈥檚 where you went wrong.”

How did Obama’s chief scientist 鈥� who was a machine learning expert 鈥� use four simple questions to help win the 2012 election?

PD: Rayid Ghani and his team of data scientists used machine learning to predict the answers to four questions for each individual swing voter, using all the data about them they could get their hands on. The questions were: How likely is he to support Obama? To show up at the polls? To respond to the campaign鈥檚 reminders to do so? And to change his mind about the election based on a conversation about a specific issue? Then, every night, they ran a program called “the Optimizer” to choose which voters to target the following day based on the results of the machine learning. In contrast, Mitt Romney鈥檚 campaign used standard polling and targeted broad demographic categories like 鈥渟uburban middle-aged woman.鈥� The result? Even though the race was close, Obama carried all the swing states but one and won the election.

How is a machine learning expert more like a farmer than a factory worker?

PD: Factory-made goods have to be assembled piece by piece, step by step, all the way from the raw materials. In contrast, crops grow on their own, with a bit of help from the farmer. Traditional computer programs are like factory-made goods; software engineers write them line by line, which is an incredibly time-consuming and error-prone process. In contrast, a machine learning expert grows programs from data in the same way that a farmer grows crops from nutrients. In our case, the seeds are learning algorithms; and big data means the soil is incredibly fertile.

What is the relationship between machine learning and artificial intelligence?

PD: The goal of artificial intelligence is to get computers to do things that in the past required human intelligence. One of those things 鈥� perhaps the hallmark of human intelligence 鈥� is the ability to learn from experience. So machine learning is a subfield of artificial intelligence, but these days it鈥檚 so successful that it鈥檚 outgrown its proud parent and has become a stand-alone field, often known by other names like data science and predictive analytics.

Lots of plot lines have been built around sentient computers that go awry or take over the world or do harm. Is this something to worry about, or are there other potential dangers?

PD: “The Terminator” scenario of an evil AI deciding to take over the world and exterminate humanity is not really something to take seriously. It鈥檚 based on confusing being intelligent with being human, when in fact the two are very different things. The robots in the movies are always humans in disguise, but real robots aren鈥檛. Computers could be infinitely intelligent and not pose any danger to us, provided we set the goals and all they do is figure out how to achieve them 鈥� like curing cancer.

On the other hand, computers can easily make serious mistakes by not understanding what we asked them to do or by not knowing enough about the real world, like the proverbial sorcerer鈥檚 apprentice. The cure for that is to make them more intelligent. People worry that computers will get too smart and take over the world, but the real problem is that they鈥檙e too stupid and they鈥檝e already taken over the world.

What has machine learning enabled university scientists and researchers to do that wouldn鈥檛 have been possible before?

PD: Machine learning is revolutionizing science by making it possible to understand much more complex phenomena than before. With it, we can apply the scientific method to vast quantities of data that no unaided human could hope to come to grips with. Biologists use machine learning to build models of the cell based on data from DNA sequencers, gene expression microarrays, and so on. Astronomers use it to automatically create catalogs of stars and galaxies from sky surveys. Physicists use it to suss out the new particles from the masses of data generated by particle colliders. Neuroscientists use it to build detailed maps of the brain, literally neuron by neuron. Social scientists use it to understand how large social networks, with millions or billions of people, behave. It鈥檚 not an exaggeration to say that machine learning and big data have ushered in a new era in science.

What is the “Master Algorithm” and how far are we from finding it?

PD: The Master Algorithm is a single algorithm capable of discovering all knowledge 鈥� past, present and future 鈥� from data. The human brain is a kind of master algorithm. So is evolution. Each has given rise to a different machine learning school, as have a number of other ideas, like symbolism. Each school has its own master algorithm: for the connectionists it鈥檚 something called backpropagation, for the symbolists it鈥檚 inverse deduction, and so on. But, as we saw, what we really need is a single algorithm that combines the capabilities of all of them. When will we find it? It鈥檚 hard to predict, because scientific progress is not linear. It could happen tomorrow, or it could take many decades. One of my fondest hopes in writing the eponymous book is that it will inspire a bright kid somewhere to come up with the key idea that we鈥檝e all been missing 鈥� and make the Master Algorithm a reality, with all the extraordinary benefits for humanity that will follow.

The key to solving many of the most important problems in business, science and technology lies in optimization 鈥� finding the values for variables that give you the highest benefit.

Whether those are which stocks to buy, which search results to return, what best predicts the outcome of the next presidential election or which amino acids to string together in a new drug to fight malaria or cancer, optimization is crucial to getting what we want. When a problem is simple, we can program a computer to solve it. When it鈥檚 too complex for that, optimization is how the computer finds the solution by itself.

91探花 machine learning researchers have developed a radically new approach to optimization, in part by borrowing a classic technique from artificial intelligence and computer science. The outlining their approach won the in July at the , the world’s largest AI conference.

In two applications that were tested experimentally, the new 91探花approach outperformed standard optimization techniques, in some instances by many orders of magnitude.

“In some ways optimization is the most important problem you鈥檝e never heard of because it turns up in all areas of science, engineering and business. But a lot of optimization problems are extremely difficult to solve because they have a huge number of variables that interact in intricate ways,” said senior author , 91探花professor of computer science and engineering.

For example, let’s say you want to teach a computer to recognize the image of a cat by examining individual pixels. Pixels that denote orange, fur or whiskers increase the chances of yes. Pointy ears or claws help confirm. Blue feathers strongly suggest no. Optimization is the art and science of weighting those variables so the machine makes the correct choice as often as possible.

The 91探花optimization algorithm, known by its acronym RDIS, progressively breaks an enormously complicated problem down into smaller, more manageable chunks 鈥� a simple idea commonly used when a problem consists of yes-or-no choices, but which had not previously been applied to numeric variables. RDIS can identify variables that, once set to specific values, break a larger problem into independent subproblems. Often, the problems are only nearly independent, but RDIS limits the error caused by treating them as fully independent.

“This approach is something that is very different than what people were doing before and it also does something magical, which is solve some problems exponentially faster. And anytime you can do that, that’s when you get a big win,” said Domingos.

The 91探花team tested RDIS against leading optimization methods in two real-world applications: determining the shape of folded proteins and accurately constructing three-dimensional objects and scenes from two-dimensional images.

For robotic arms to perform surgeries or to prevent self-driving cars from crashing, computers must accurately map the two-dimensional camera images that serve as the robot’s “eyes” into realistic three-dimensional spaces. The 91探花team’s optimization approach, on average, performed that task between 100,000 and 10 billion times more accurately than previous methods.

“You need to minimize the reconstruction error 鈥� the difference between what the algorithm predicts and what the image actually shows. Our algorithm did this very well,” said lead author , a 91探花computer science and engineering doctoral student. “This is crucial for a self-driving car 鈥� in order to safely navigate its environment, it first needs to be able to determine what objects are close to it, where the road is, and where the other cars are.”

The researchers also tested RDIS against other optimization techniques for protein folding. A string of amino acids 鈥� protein building blocks that can occur in millions of different patterns 鈥� will generally fold and twist into a shape with the lowest energy. That configuration is important because it dictates how the protein interacts with viruses or invading cells or other parts of the immune system.

Enabling computers to accurately predict how proteins will fold can greatly speed the process of designing effective drugs. The 91探花team found its new optimization technique resulted in much lower-energy protein shapes than alternative methods, particularly for larger proteins.

That’s because current optimization approaches fail as problems get more complex, Domingos said. Solving optimization problems is like being left blindfolded at the top of a hill and asked to walk to the ocean. One way to do that is to judge where to go by feeling around with your foot and taking one step a time in the steepest downward direction. That works if there’s only one hill. But if you’re at the top of the Himalayas, you鈥檒l quickly get stuck because there are thousands of peaks and foothills and flat parts. That’s essentially what happens with current optimization algorithms.

“If you’re lucky, maybe you’ll wind up in the sea but more likely you’ll wind up in a valley or a lake,” said Domingos. “If you could see the whole landscape, you’d say ‘oh, here’s where I have to go,’ but the problem is you can’t see everywhere and neither can today’s algorithms.”

To address those deficiencies, the 91探花duo figured out how to apply decomposition techniques, commonly used by artificial intelligence researchers and puzzle solvers, to continuous optimization problems, which have historically been the bailiwick of engineers, applied mathematicians and physicists.

The next steps, researchers say, are to test the algorithm’s performance on new and different applications. “This can be applied to pretty much any machine learning problem, but that’s not to say it’s going to be good for every machine learning problem,” said Domingos. “That’s what we have to work on and find out.”

The research was partly funded by the U.S. Office of Naval Research, Army Research Office, and Defense Advanced Research Projects Agency.

For more information, contact Pedro Domingos at 206-543-4229 or pedrod@cs.washington.edu.