This rough crew rocks the world of earthquake prediction

Jeff J. Mitchell / Getty Images / Wired

The Holy Grail of Seismology is forecasting earthquakes. Unpredictability, which gives the probability that a large earthquake will occur in a region over a period of years or decades, but in reality it predicts exactly when and where the earthquake will strike itself, and how large it will be.

Seismologists will tell you unequivocally that anyone claiming that they can predict future earthquakes is a hoax. But this does not mean that predicting earthquakes is impossible. Scientists have invested an enormous amount in achieving this sacred goal, but so far they have failed.

In December 2018, a coalition of researchers decided to try something new. They announced an online competition, open to anyone, in which participants had to anticipate future earthquakes produced by a device resembling a vice in the laboratory. has evolved? They had to design their rudimentary AI to make predictions.

Thousands of people from all over the world threw their hats into the ring. As reported in an article published in the Proceedings of the National Academy of Sciences earlier this month, the winning teams were able to come up with sets of code that predict the timing of future laboratory earthquakes with amazing accuracy.

It remains unclear how this applies to a fault zone in real life. But the promise of these new machine learning models suggests that predicting earthquakes is not a far-fetched dream, but a plausible possibility. And with none of the winners having a background in seismology, this contest demonstrates the benefits of searching for an extremely wide network to find hidden talents – the kind that might one day save millions of lives.

Earthquakes have been around forever, but seismology is still in its youth. The kind of quantitative data we’re all used to today – the magnitudes, the intensity of vibration and everything – hasn’t been around for very long. It was only in the twentieth century that the dawn of robotic seismology occurred. Certainly, ancient texts and indigenous knowledge describing earthquakes help fill the deficit in historical data, and some countries, such as Japan, have a much longer written record of tremors and earthquakes.

But to properly study earthquakes, you have to document them as soon as they happen. Far from the invention of time travel, seismologists – at least compared to other Earth scientists – will still suffer from a lack of data.

“Our seismic record is rather short,” says Casey Aderhold, a seismologist at IRIS, a consortium of seismologists. This means that despite tremendous advances in the past century, our understanding of the physics that drive earthquakes is a bit vague and somewhat theoretical.

However, there is just enough data to look into the future – to an extent. Organizations like the U.S. Geological Survey, using knowledge of past earthquakes and current geophysical information, say that, for example, there is a 20 percent chance that the San Francisco Bay Area will experience a magnitude 7.5 earthquake in the next 30 years.

They can also predict aftershocks, the earthquakes following the main shock, the strongest earthquake in the sequence. When a major earthquake occurs (possibly a major shock), a series of equations and calculations – those derived from statistical models and plausible assumptions about how earthquakes work – produce a time window (for example, a week) in which there are an approximate number of aftershocks with magnitude An approximation is likely to happen.

These accounts are very powerful. But they are retrospective and interactive, and cannot be used to predict the major shocks, stars of performances that could throw humans and their homes. The problem is the lack of precursors: seismologists have not yet definitively identified the signals that precede major shocks that clearly indicate their appearance and size.

When it comes to predicting earthquakes, “There is a long history of […] Let’s call it a recession, on the subject with a little serious success, “says Zachary Ross, a seismologist at Caltech who was not involved in the work.” Compared to the kind of progress people have made in the climate over the past several decades, which is amazing, The current technologies [earthquake] The prediction appears to be somewhat saturated with what they seem to be able to do. “

Enter, machine learning. Crudely, this describes the ability of a computer code to assimilate data, identify patterns, make choices or predictions, and then learn from its mistakes to correct itself – all without much human intervention. For seismologists, it is a novelty. They have only started discussing her potential and their work with her at major scientific conferences in the past few years.

However it is already used in the real world. Project Ross was part of the machine learning used to find millions of earthquakes buried in earthquake records in Southern California. After being exposed to blocks of seismic data, their software was able to quickly distinguish between random rumbling and the real grumbling of earthquakes, the kind that humans cannot imagine.

Paul Johnson, a geophysicist at Los Alamos National Laboratory in New Mexico and lead author of the new study, thinks machine learning may be able to help predict earthquakes. Instead of using equations designed around human understanding of seismology, these codes will start over, consuming data and using that data alone to make predictions – and removing potentially wrong human assumptions from the mix.

A previous study used an in vitro synthetic earthquake-making machine. The steel blocks were confined by a block of crack pits, a rock usually found in natural faults. The blocks were mechanically moved, pushed, crushed, and retracted the block. If the block cracks and jolts forward, voila, it caused an earthquake.

Johnson and colleagues wondered whether the machine-learning model, which was provided with a stream of data about the dimensions and properties of the “rift,” the stress and pressure the system undergoes and the resulting earthquakes, could predict future earthquakes. “Very quickly, a model was developed by some talented young people,” he says – a team of materials scientists and mathematicians. Within months of preparing the device, their software could predict with great accuracy when future earthquakes will occur.

“It was just an advertisement,” Johnson says. “It looked like magic at first.” The model was able to identify a small number of energy signatures in the seismic data that allowed it to know that the fault was about to fail.

But if this is a video game then it is set to “easy” difficulty. Earthquakes were periodic, meaning they occurred fairly regularly. Can machine learning deal with irregular earthquakes, the kind you see in the wild?

In the nineteenth century, scientific contests between competing researchers were commonplace, as white men who often came from money cast a shadow over each other as they tried to prove the prowess of their theories in explaining reality in action. In recent years, scientific competitions have become more focused on devising technological solutions to problems, and usually involve group efforts.

Johnson and his colleagues did not realize why this had not been applied to machine learning. They turned to Kaggle, a platform that machine learning advocates use to share research. In the past, it was used to host competitions, including one where people tried to spot dark matter. In late 2018, they got rid of the challenge: their laboratory malfunctions machine was going to simulate earthquakes in the first half of 2019, and they wanted machine learning models assigned to people to predict when they would happen. The top five teams will share a $ 50,000 (£ 36,000) rating.

Ultimately, 4,521 teams comprising 5,450 people signed up. They come from a dizzying array of backgrounds, from mobile gamers to cartoonists, and from insurance salespeople to those studying the electrical signals produced by hearts and brains. The teams first built their models using the training data: seismic signals, earthquake timing, shear pressure the fault was experiencing, etc. They then submitted up to two daily models to the arbitrators, who put them in their paces in an effort to predict device earthquakes while receiving only seismic data. Results are given based on how accurate these predictions are.

The top five teams – GloryorDeath, Reza, Character Ranking, JunKoda and The Zoo – have become masters of earthquake prediction. The first team, The Zoo, was a team of eight from the US and Europe, quite a mix of acquaintances and strangers. Despite the potential for disorganized chaos, they managed to take first place thanks to some clever hacks.

The first was to build their model using not only training data, but also with test data, thus making the test a learning experience rather than just a grueling challenge. In some circles, Johnson says, this could be considered fraud. But this is how machine learning will actually work: It will learn from training data sets and from its experiences with real earthquakes.

The second winning feature was not as intuitive as it was inspired: It fed a noise into the data stream. Noise – caused by traffic, wind, ocean, animals, or people walking – is a bane for seismologists, who need to filter it to hear earthquakes. It’s not actually clear why their model was made more accurate. “Some of the things people do just work, and you don’t necessarily understand why,” Johnson says. One possibility is that you are simply giving the models more data to practice with and learn from. Practice makes perfect, after all.

Remarkably, none of the winners had a background in seismology. Are the judges affected by this revelation? Not at all, says Laura Perak Nollet, an astronomer and physicist at Purdue University and co-author of the study. “For us, it was a very exciting experience.” Aderhold says this real progress made in a highly collaborative framework will only help convey the myth of a disturbing lonely genius into history.

Machine learning has already demonstrated predictive capabilities in the Cascadia subduction zone in the Pacific Northwest. After hearing 12 years of seismic soundtracks popping up from a very gradual fault movement called “slow slip”, I was able to find patterns in the loud parts that anticipated the next slow slip, such as knowing when the pitch was about to drop in a song.

The next step is to try it out on relatively quiet mistakes that will someday violently fail and shake the ground. “We’re really in the midst of that right now, and we don’t know what the outcome will be,” Johnson says.

Laboratory earthquakes are simplified versions of bona fide errors. Success in this competition then doesn’t mean that machine learning has unveiled the Holy Grail. But it is clear that resolving the issue will not be easy, if success is possible. “It remains to be seen if there will be progress in our ability to predict real earthquakes using machine learning,” Ross says.

But not knowing if there is a way to predict earthquakes will not stop seismologists and their new colleagues around the world from trying. Aderhold invokes a Douglas Adams quote from his wonderful authorship, The Traveler’s Guide to the Galaxy: “There is the art of flying, or rather a talent. The talent is in learning how to throw yourself to the ground and miss.” Seismologists hit the Earth a lot in their quest, but they still dreamed of the journey.

More great stories from WIRED

? A rebellious physicist with a neat solution to the quantum puzzle

? Google is rewriting the web. Here’s the effect that Chrome’s plan to kill cookies will have

? As more Covid-19 variants emerge, attention has shifted to the N95 and FFP2 face masks

? Listen to The WIRED Podcast, The Week in Science, Technology and Culture, presented every Friday

? Follow WIRED on Twitter, Instagram, Facebook and LinkedIn

Get WIRED Daily, your candid summary of all the biggest stories in technology, business and science. In your inbox every day of the week at 12 pm UK time.

By entering your email address, you agree to our privacy policy

Thank you so much. You have successfully followed our post. You will hear from us soon.

Sorry, you entered an invalid email address. Please update and try again.

What Are The Main Benefits Of Comparing Car Insurance Quotes Online

LOS ANGELES, CA / ACCESSWIRE / June 24, 2020, / Compare-autoinsurance.Org has launched a new blog post that presents the main benefits of comparing multiple car insurance quotes. For more info and free online quotes, please visit https://compare-autoinsurance.Org/the-advantages-of-comparing-prices-with-car-insurance-quotes-online/ The modern society has numerous technological advantages. One important advantage is the speed at which information is sent and received. With the help of the internet, the shopping habits of many persons have drastically changed. The car insurance industry hasn't remained untouched by these changes. On the internet, drivers can compare insurance prices and find out which sellers have the best offers. View photos The advantages of comparing online car insurance quotes are the following: Online quotes can be obtained from anywhere and at any time. Unlike physical insurance agencies, websites don't have a specific schedule and they are available at any time. Drivers that have busy working schedules, can compare quotes from anywhere and at any time, even at midnight. Multiple choices. Almost all insurance providers, no matter if they are well-known brands or just local insurers, have an online presence. Online quotes will allow policyholders the chance to discover multiple insurance companies and check their prices. Drivers are no longer required to get quotes from just a few known insurance companies. Also, local and regional insurers can provide lower insurance rates for the same services. Accurate insurance estimates. Online quotes can only be accurate if the customers provide accurate and real info about their car models and driving history. Lying about past driving incidents can make the price estimates to be lower, but when dealing with an insurance company lying to them is useless. Usually, insurance companies will do research about a potential customer before granting him coverage. Online quotes can be sorted easily. Although drivers are recommended to not choose a policy just based on its price, drivers can easily sort quotes by insurance price. Using brokerage websites will allow drivers to get quotes from multiple insurers, thus making the comparison faster and easier. For additional info, money-saving tips, and free car insurance quotes, visit https://compare-autoinsurance.Org/ Compare-autoinsurance.Org is an online provider of life, home, health, and auto insurance quotes. This website is unique because it does not simply stick to one kind of insurance provider, but brings the clients the best deals from many different online insurance carriers. In this way, clients have access to offers from multiple carriers all in one place: this website. On this site, customers have access to quotes for insurance plans from various agencies, such as local or nationwide agencies, brand names insurance companies, etc. "Online quotes can easily help drivers obtain better car insurance deals. All they have to do is to complete an online form with accurate and real info, then compare prices", said Russell Rabichev, Marketing Director of Internet Marketing Company. CONTACT: Company Name: Internet Marketing CompanyPerson for contact Name: Gurgu CPhone Number: (818) 359-3898Email: [email protected]: https://compare-autoinsurance.Org/ SOURCE: Compare-autoinsurance.Org View source version on accesswire.Com:https://www.Accesswire.Com/595055/What-Are-The-Main-Benefits-Of-Comparing-Car-Insurance-Quotes-Online View photos