Connect with us

FACEBOOK

Inside the Instagram AI that fills Explore with fresh, juicy content

Published

on

Instagram has posted an article describing the behind-the-scenes machinery that fills the Explore tab in Instagram with new, interesting stuff every time you open it. It’s a bit technical, so here are five takeaways.

Even Instagram and Facebook have limited resources

Unlike the feed, which some still would prefer was simply chronological, the Explore tab needs to be algorithmically driven. But understanding what’s happening on an image-based social network and recommending new content to people is a problem that’s exactly as hard as you make it.

If these companies had infinite processing power and time, they’d probably come at the question of Explore a bit differently. But as it is they need to serve hundreds of millions of people on short notice and with merely enormous computing resources. I think they put this at the top of the post so people don’t wonder why they’re cutting corners.

It’s also easier to experiment and iterate when you can change stuff and see results quickly, they point out.

It’s all about the account, not the post

So much is posted to Instagram that it would be pretty much impossible to keep track of every photo individually, for recommendation purposes anyway. It’s simpler and more efficient to track accounts, since accounts tend to have themes or topics, from a broader one like “travel” to something highly specific, like especially round seals.

While liking one post from an account doesn’t necessarily mean you’ll like everything else from that account, it is a good indicator that you’re at least interested in the theme of that account. Even if it was this particular post of this particular cat that you wanted to heart because it reminds you of old Mittens, if you’re liking pictures from an account that mostly posts cats, that’s valuable information.

Complex habits inform the algorithm

Notably it isn’t just image features that Instagram uses to figure out what accounts are topically linked, though of course that kind of thing can be detected too. They also use your behavior.

For instance, when you like several posts in a row, they’re more likely to be linked in some way even if Instagram’s algorithms can’t quite see it:

If an individual interacts with a sequence of accounts in the same session, it’s more likely to be topically coherent compared with a random sequence of accounts from the diverse range of Instagram accounts. This helps us identify topically similar accounts.

People just tend to look into stuff that way, going from one travel-focused account to the next, or focusing on animals because they need a pick-me up. All that information gets sucked up by the algorithm and inspected for relevance. Of course deliberate actions like “see fewer posts like this” and blocking accounts has a lot of weight as well.

From “seed accounts” to a top 25

The process of getting from a couple billion posts to just two dozen can be pretty difficult, but you can cut the problem down to manageable size by limiting the Explore tab to accounts linked in some way to accounts the user has already liked or saved posts from. These are called “seed accounts” because everything else in the process really grows out of them.

Because of how the machine learning system represents accounts and their topics inside itself, it’s super easy for it to find a couple hundred similar accounts.

Imagine if you know someone likes a particular reddish-orange marble and you need to find some more like it. If you just dip your hand into a sack of marbles you’re unlikely to find one quickly. Even if you pour them out on the floor you’ll still have to hunt around for a bit. But if you’ve already organized them by color, all you have to do is reach into the general vicinity of the marble they like and you’re almost guaranteed to pick a winner.

The machine learning model does that by giving all these accounts a sort of location in a virtual space, and the closer two are in that space, the closer they are topically.

So the really hard part of paring down a set of billions to a set of hundreds is basically already accomplished by the way the accounts are classified.

From there Instagram does three passes with neural networks of increasing complexity.

First, slightly confusingly, is a simpler, combined version of the next two processes, which takes it from 500 to 150 accounts. This is a little weird, but think about it this way: This neural network has seen steps 2 and 3 happen many times and has a pretty good idea of what they do. Sort of like if you’d seen cookies get made enough times that you could guess at a recipe. You’d probably get close, but you also wouldn’t want to publish it to like a hundred million people. So this step just gets the obvious stuff right.

Second is a computationally cheap neural network that uses way more signals than the simple topical similarity mentioned above. Here’s where your individual likes come into play, as well as the deeper data about accounts. You like travel, sure, but in particular you like couples traveling — both things the marble-sorting algorithm above can help with. Other parameters, like a post’s general popularity, or actually its being different from the other posts in the mix, figure in as well. That skims another 100 off the top, leaving 50.

Third is a computationally expensive version of the above, which does another pass on those 50 and cuts them in half, basically by looking closer and taking the time to include, perhaps, a thousand data points each rather than a hundred.

I guess that was kind of long for a “takeaway.” Don’t worry, the next one is quick.

And of course, no 🍑

“We want to make sure the content we recommend is both safe and appropriate for a global community of many ages on Explore,” they write. “Using a variety of signals, we filter out content we can identify as not being eligible to be recommended.”

So now you know why you don’t get any of that in Explore.

TechCrunch

FACEBOOK

Stock and Share Market News, Economy and Finance News, Sensex, Nifty, Global Market, NSE, BSE Live IPO News

Published

on

Stock and Share Market News, Economy and Finance News, Sensex, Nifty, Global Market, NSE, BSE Live IPO News


By



Source link

Continue Reading

FACEBOOK

Amazon’s advertising business grew 19%, unlike Google, Meta

Published

on

Amazon's advertising business grew 19%, unlike Google, Meta

The regulator was concerned with Amazon’s dual role as both a marketplace and a competitor to merchants selling on its platform.

Nathan Stirk | Getty Images

Amazon’s advertising business continues to grow despite a general slowdown in digital advertising, which has hurt companies like Google parent Alphabet, Facebook parent Meta and Snap.

The online retail giant’s advertising services unit brought in $11.6 billion in sales for the fourth quarter, representing a 19% year-over-year increase, according to its earnings report Thursday.

Although Amazon’s advertising unit still constitutes a small fraction of the $149.2 billion in revenue the company recorded in its fourth quarter, it represents a fast-growing area that analysts believe could be a crucial player in the digital advertising market.

Indeed, while investors were pleased that Meta is cutting costs, sales in the company’s fourth quarter dropped 4% year-over-year to $32.17 billion.

Meta executives explained on Wednesday during a call with analysts that they don’t see an immediate rebound in the digital advertising market coming anytime soon. Susan Li, the company’s chief financial officer, said “Consistent with our expectations, Q4 revenue remained under pressure from weak advertising demand, which we believe continues to be impacted by the uncertain and volatile macroeconomic landscape.”

Meanwhile, Alphabet on Thursday reported fourth quarter advertising revenues of $59.04 billion, a slowdown from $61.24 billion in the year-ago quarter.

Alphabet’s YouTube advertising unit, which faces competition from TikTok, brought in $7.96 billion in the fourth quarter, representing an 8% drop from a year ago.

Tech companies that are powered by digital advertising have been under pressure from several factors, including a tough economy, increased competition from TikTok, and the lingering effects of Apple‘s 2021 iOS privacy update.

The latest Insider Intelligence survey of digital advertising revenue share worldwide revealed that Amazon now holds 7.3% of the overall online ad market, trailing Alphabet‘s Google, and Meta-owned Facebook and Instagram, which respectively have 28.8%, 11.4%, and 9.1%, of the digital ad market.

Source link

Continue Reading

FACEBOOK

FTC sets its sights on the health data market

Published

on

FTC sets its sights on the health data market

With Carmen Paun

AN FTC FIRST — The Federal Trade Commission is cracking down on companies sharing health data in new ways that could have implications for online business models, POLITICO’s Ruth Reader reports.

The FTC said Wednesday it had reached an agreement with GoodRx on a fine and remedies after the prescription discount site and telehealth provider shared customers’ health data with Google, Facebook and other third parties.

It’s the commission’s first enforcement of its 2009 Health Breach Notification Rule. Pending a court’s agreement, the decision could upend business models that rely on selling or using the data.

In the agreement with GoodRx, the agency filed a proposed order to levy a $1.5 million fine and enforce the remedies with the federal court in the northern district of California, which still must approve the agreement.

The FTC said GoodRx was unfair in its handling of customer data — alleging the company falsely claimed it complied with HIPAA and also shared information when it pledged not to. The company also had no internal processes to protect, or limit third-party access to, consumer health data, the commission said.

Though GoodRx has agreed to settle, it didn’t admit wrongdoing. The company also said it didn’t believe the FTC action would materially impact the business.

“We believe this is a novel application of the Health Breath Notification Rule by the FTC. We used Facebook tracking pixels to advertise in a way that we feel was compliant with regulations and that remains common practice for many websites,” the company said in a statement.

Still, the FTC is signaling heightened interest in the issue, data privacy experts say.

“What they’re doing is sending a warning shot across the digital bow of the online advertising industry saying, ‘Hey, these things are unfair, we’re watching, and you should not be using this health information in the way it’s being used,’” said Jeff Chester, executive director of the Center for Digital Democracy, a Washington, D.C.-based nonprofit that advocates for digital privacy and consumer protections online.

WELCOME TO THURSDAY PULSE. One bill to keep an eye on: a proposition to make roasted chile New Mexico’s official state aroma. What should your state’s official aroma be? Send ideas — and health news — to [email protected] and [email protected].

TODAY ON OUR PULSE CHECK PODCAST, Erin Schumaker talks with Megan Messerly about the millions of Americans who were allowed to remain covered by Medicaid during the pandemic and what could happen now that Congress has given states the go-ahead to reevaluate who’s still eligible for those health insurance benefits.

AMERICA DOESN’T HAVE THE CAPACITY TO IDENTIFY PANDEMIC ORIGINS, EXPERTS TELL CONGRESS — The United States doesn’t have the combination of scientific research, access to samples databases, domestic operational plans and international partnerships that can reliably identify the source of disease outbreaks such as the coronavirus pandemic, five experts told a House Energy and Commerce subcommittee hearing Wednesday, Carmen reports.

Using genetic sequencing and analyzing blood samples stored in databases are technologies that have proven useful in detecting the origin of other diseases in the past, but a lack of access to such databases hinders origin investigation, Karen Howard, the acting chief scientist at the Government Accountability Office told the subcommittee on oversight and investigations.

International agreements should be developed to standardize sample databases that could help in researching a virus’ origin, she said, adding that the U.S. should also develop a detailed national strategy for investigating a pandemic’s beginning, she said.

One single office in the U.S. government should coordinate the work of several agencies in identifying where an outbreak started, added Tom Inglesby, the director of the Center for Health Security at the Johns Hopkins Bloomberg School of Public Health and a former adviser on the White House Covid-19 Response Team.

Why it matters: Inglesby said the ability to investigate a viral outbreak could be a form of deterrence against enemies who would want to use biological weapons against the U.S.

The hearing was the first for this Congress that focused on Covid’s origins, an issue that the Republican House majority has made a priority. Most questions from the Republican subcommittee members focused on whether it could be demonstrated that the coronavirus originated at the Wuhan virology lab.

The health experts testifying said the current data doesn’t clearly trace the virus to the Wuhan lab, but several studies link it to a live animal market in that Chinese city.

Michael Imperiale, a professor of microbiology and immunology at the University of Michigan, warned against politicizing the debate and discouraged scientists from getting involved in such research. Some of his colleagues studying viruses with pandemic potential, he said, have received death threats from people who mistrust the researcher’s work, suspecting them of deliberately engineering viruses to become more transmissible or dangerous.

“We must be careful not to throw sand in the gear that slows our progress, dissuades our scientists or discourages our young people from being a part of our scientific system,” he told lawmakers.

WYDEN WANTS INFO ON IRA REBATES — Sen. Ron Wyden (D-Ore.) sent a letter Wednesday to CMS Administrator Chiquita Brooks-LaSure asking for details about the Medicare Part D and Part B inflation rebate provisions included in the Inflation Reduction Act.

The information requested includes a timeline for implementing rebates, an explanation of how those rebates will be calculated and a plan to promptly penalize companies that increase prices faster than the inflation rate.

ONCDP TO THE CABINET? A bipartisan group of 55 lawmakers asked President Joe Biden in a letter Wednesday to add the director of the Office of National Drug Control Policy to a Cabinet-level position.

The lawmakers wrote that, amid the opioid epidemic, the president should announce the change at next week’s State of the Union address and push ending the crisis as a top priority.

IN CASE OF DEFAULT — The largest House Republican caucus worked on a list of ideas for fiscal reform, including an item on Medicare, POLITICO’s Caitlin Emma and Olivia Beavers report.

Though Speaker Kevin McCarthy said earlier this week that Medicare and Social Security were off the table for cuts, the group is considering a way to continue payments to beneficiaries should the U.S. default on its debt.

FDA EMPLOYEES WON’T BE FIRED OVER FORMULA CRISIS — As the FDA looks to major reforms in the wake of the infant formula crisis, the agency’s commissioner said employees won’t be fired or reassigned in the changes, POLITICO’s Meredith Lee Hill reports.

The announcement came as Commissioner Robert Califf rolled out his “new, transformative vision” of the main agency tasked with overseeing food safety in the U.S., though he didn’t include specific plans to address breakdowns around infant formula.

Still, Califf pointed to some past “leadership changes.” His remarks come just days after senior FDA foods official Frank Yiannas’ resignation last week. In his resignation letter, Yiannas called for structural reforms in the troubled division.

“But the short answer is no one’s going to be reassigned or fired because of the infant formula situation,” Califf told reporters.

PANDEMIC PREP DEAL DETAILS — The World Health Organization shared plans for an international agreement aimed at improving pandemic preparations, Carmen reports.

The plan lays out ideas to avoid the failures from the Covid-19 pandemic, such as inequitable vaccine distribution.

The proposal would require countries to allow WHO rapid-response teams access to their territories to assess and support efforts to combat emerging outbreaks — after China didn’t grant fast access to international experts to the Wuhan virology lab at the pandemic’s outset.

The draft also demands that countries support temporary waivers of intellectual property rights on those products and requires manufacturers that received public funding for their development to waive their rights. That sort of provision, hotly contested through the Covid era, will likely be fought by pharmaceutical companies.

Governments will start negotiations on the agreement at a meeting later this month, with discussions continuing for the next year.

FIRST IN PULSE: Andrea Harris, previously chief of staff to Rep. Lauren Underwood (D-Ill.) and two HHS assistant secretaries, will join Protect Our Care as director of policy programs.

Jean Accius is now president and CEO of Creating Healthier Communities. He previously was SVP of global thought leadership for AARP.

The New York Times reports that vaccine makers kept well over $1 billion in prepayments for Covid shots for developing countries.

Kaiser Health News writes about nursing home owners funneling cash out of facilities during the pandemic.

The Washington Post reports on research about the cancer risk associated with ultra-processed food.

Source link

Continue Reading

Trending

en_USEnglish