Mediaboss Marketing

Covering COVID-19 is a daily Poynter briefing of story ideas about the coronavirus and other timely topics for journalists, written by senior faculty Al Tompkins. Sign up here to have it delivered to your inbox every weekday morning.

This does not help. Because Nicki Minaj sent a tweet to her 22.6 million Twitter followers, we now shall spend a few paragraphs telling unvaccinated young males that there is no evidence that the COVID-19 vaccine will harm their testicles or fertility.

The Centers for Disease Control and Prevention reiterated that there are no known side effects from the COVID-19 vaccine like swollen testicles or fertility issues. None. The CDC says:

Currently no evidence shows that any vaccines, including COVID-19 vaccines, cause male fertility problems. A recent small study of 45 healthy men who received an mRNA COVID-19 vaccine (i.e., Pfizer-BioNTech or Moderna) looked at sperm characteristics, like quantity and movement, before and after vaccination. Researchers found no significant changes in these sperm characteristics after vaccination.

Fever from illness has been associated with short-term decrease in sperm production in healthy men. Although fever can be a side effect of COVID-19 vaccination, there is no current evidence that fever after COVID-vaccination affects sperm production.

Twitter made a big announcement in March that it would attach warning labels to tweets that contain false information about COVID-19. Twitter did not block this post or this poster.

Later, Minaj added that she is still uncertain about taking the vaccine.

She then said she was leaning toward getting vaccinated.

Ill add one more piece of information to this conversation. While there is no proof that the vaccination is connected to erectile dysfunction or male infertility, there is evidence that getting the virus can cause those problems.

Senator Rand Paul of Ky., addresses the audience at the Kentucky Farm Bureau Ham Breakfast at the Kentucky State Fair in Louisville, Ky., Thursday, Aug. 26, 2021. (AP Photo/Timothy D. Easley)

Three medical groups the American Board of Family Medicine, the American Board of Internal Medicine and the American Board of Pediatrics are warning physicians who spread false information about COVID-19 that they could lose their licenses. But, so far, it is all talk and no action.

We also want all physicians certified by our boards to know that such unethical or unprofessional conduct may prompt their respective board to take action that could put their certification at risk, the boards wrote.

Not long ago, the Federation of State Medical Boards warned that physicians who spread COVID-19 rumors could lose their state licenses. Now, the professional associations say they could lose their board credentials.

The Mississippi State Board of Medical Licensure adopted a new policy last week saying physicians have an ethical and professional responsibility to act in the best interest of their patients and, Spreading inaccurate COVID-19 vaccine information contradicts that responsibility, threatens to further erode public trust in the medical profession and puts all patients at risk. And, the policy adds, Physicians must understand that actions online and content posted can affect their reputation, have consequences for their medical careers, and undermine public trust in the medical profession.

The Tribune News Service put the story in perspective:

Experts fear that a deepening distrust of expertise among many Americans, the reach offered by social media, and national politicians who promote bogus covid-19 theories are creating a welcoming environment for doctors and nurses who traffic in dangerous falsehoods that can be both alluring and bewildering given how quickly knowledge of covid-19 has evolved.

When are we talking about honest differences of opinion and when are we talking about a flagrant disregard of standards of care? asked Richard Baron, a doctor and head of the Philadelphia-based American Board of Internal Medicine. With respect to some of the behavior were seeing it really is in contravention of pretty solid professional science.

MedPage searched for an example of a tough-talking state taking action against a nonsense-spewing physician and could not find one. Not one.

Despite a national call to sanction doctors who spread COVID-19 misinformation, a MedPage Today investigation found that not one of 20 physicians whove peddled such falsehoods has been disciplined by their state licensing agency for doing so.

Thats not to say that complaints havent been filed, or that investigations havent been launched. These elements are confidential in most states, including the 10 contacted by MedPage Today who license the physicians.

But that means physicians who have advanced false COVID information including Simone Gold, MD, JD; Scott Atlas, MD; Joseph Mercola, DO; Lee Merritt, MD; Sherri Tenpenny, DO; and Stella Immanuel, MD are free to continue to misinform their patients and the public, even as the Delta variant surges.

Our statement is a reminder to physicians that words have consequences and during a public health emergency like COVID-19, those words can mean life or death for patients, Joe Knickrehm, vice president of communications for the Federation of State Medical Boards (FSMB), told MedPage Today via email.

Sen. Rand Paul, who is an ophthalmologist, also opposes COVID-19 vaccines and claims cloth masks are not useful in controlling the spread of the virus. YouTube suspended him from its platform for making false statements. And despite false online rumors that the American Medical Association took action against him, Paul is still licensed to practice medicine. For one thing, the AMA does not license physicians. States do.

One of the ways that we have come to understand the severity of the pandemic is to see how many people are hospitalized with COVID-19. But researchers have wondered if that is such a good measure since the data shows us how many are hospitalized but not how severely ill those patients are. The Atlantic explains what they found:

The study found that from March 2020 through early January 2021before vaccination was widespread, and before the Delta variant had arrivedthe proportion of patients with mild or asymptomatic disease was 36 percent. From mid-January through the end of June 2021, however, that number rose to 48 percent. In other words, the study suggests that roughly half of all the hospitalized patients showing up on COVID-data dashboards in 2021 may have been admitted for another reason entirely or had only a mild presentation of disease.

This increase was even bigger for vaccinated hospital patients, of whom 57 percent had mild or asymptomatic disease. But unvaccinated patients have also been showing up with less severe symptoms, on average, than earlier in the pandemic: The study found that 45 percent of their cases were mild or asymptomatic since January 21. According to Shira Doron, an infectious-disease physician and hospital epidemiologist at Tufts Medical Center, in Boston, and one of the studys co-authors, the latter finding may be explained by the fact that unvaccinated patients in the vaccine era tend to be a younger cohort who are less vulnerable to COVID and may be more likely to have been infected in the past.

A little disclaimer about this study. First, it was done on Veterans Affairs patients, meaning it includes relatively few women and no children. Second, the data comes from patients who were infected before the delta variant was widespread, so it is possible that the wave hospitalized now are sicker than those in the test.

The most we can pull from this is that hospitalizations, taken alone, may not tell us as much as they might seem to be saying.

Just as the committee that advises the Food and Drug Administration about vaccine safety and effectiveness is about to meet, two outgoing FDA vaccine regulators are saying that there is not a compelling reason (yet) to administer COVID-19 booster shots to the general public.

The Lancet, a respected medical journal, just published the paper by Marion Gruber and Phil Krause, who have been leading the FDAs vaccine approval process but announced they will be leaving the FDA soon. The key quote from the paper is, Current evidence does not, therefore, appear to show a need for boosting in the general population, in which efficacy against severe disease remains high.

The paper says that everyone may need a COVID-19 vaccination booster in the future, but for now, the vaccine is doing a good job.

There is general agreement that people with compromised immune systems, such as cancer patients, would benefit from booster vaccinations. There is also some agreement that a booster would benefit senior citizens. But that is where the agreement ends. Heres a good background article from PolitiFact.

In Colorado, educators say they could face arrest and criminal prosecution if they fail to wear masks and enforce mask mandates. The local district attorney in Littleton says such charges are possible, but no complaints or charges are being considered at the moment.

A student wears hand sanitizer at Tussahaw Elementary school on Wednesday, Aug. 4, 2021, in McDonough, Ga. (AP Photo/Brynn Anderson)

KDKA Pittsburgh gives us something new to worry about:

Dr. Michael Lynch, head of the Pittsburgh Poison Center, says theres been an increase of kids ingesting hand sanitizer.

In Pennsylvania, in the last 18 months, more or less since the start of the pandemic, weve seen a 56% increase in hand sanitizer exposure cases compared to the 18 months prior to that, he said.

Dr. Lynch says that amounts to about 2,300 cases, most under the age of 5, and about 6% of those kids needed to go to the hospital.

Hand sanitizers will typically have 70% or so ethanol, which means theyre about 140 proof, Dr. Lynch said. Even a small amount can be enough to get intoxicated.

Make of this what you will. TV Newser reports:

Fox Corp. human resources chief Kevin Lord sent out a memo to staff Tuesday saying more than 90% of our full-time employees reported that they are fully vaccinated after the company mandated everyone report their vaccination status.

Lord added that the company will soon introduce daily COVID testing for the small group of employees who are not vaccinated or have not provided their vaccination status.

Consulting firm KPMG surveyed more than 100 retail executives and heard that there are going to be lots of warehouse, delivery and retail jobs open this holiday shopping season. UPS alone is planning to hire 100,000 part-time and full-time seasonal workers. The U.S. Postal Service says it will add 40,000 workers this fall. Discount retailers like Kohls and Michaels also say they will hire tens of thousands of seasonal workers because they expect people to be in a spending mood this year.

Crab cake potato stacks made with Old Bay Seasoning are seen on Nov. 9, 2010, in Concord, N.H. (AP Photo/Larry Crowe)

Axios Charlotte reports:

A national crabmeat shortage has caused a 40-50% drop in business for the owners of Lulus Maryland-Style Chicken and Seafood. Every week it seems, they encounter another angry customer whos come to bite into the jumbo lump hype, only to be disappointed.

Ive had to say this so many times: Were not turning them away because we dont want to give it to them, Jay tells me. Were turning the away because we dont have it.

NPR says there are lots of reasons for the shortage that has driven prices up 50%, including fewer fishermen, a shortage of truck drivers and imports being interrupted by shipping problems.

Some restaurants are controlling prices by shrinking the size of crab cakes. If you are still paying the same price now as you were a year ago, Axios says youre probably eating way more cake than crab.

Old restaurant joke: Why do crabs never give waiters a tip? Because theyre shellfish.

Im here all week.

Well be back tomorrow with a new edition of Covering COVID-19. Are you subscribed? Sign up hereto get it delivered right to your inbox.

See the original post:
The problems that Nicki Minaj caused for the vaccination effort Poynter - Poynter

With Nicholas Wu.

HOW DO YOU SOLVE A PROBLEM LIKE Sens. Joe Manchin (D-W.Va.) and Kyrsten Sinema (D-Ariz.) are linchpins for Democrats ability to move their reconciliation spending package but aren't on board with the $3.5 trillion price tag and some key policies their party wants to champion, from climate provisions to child tax credits. But other Democrats aren't ready to roast their holdout colleagues. The political reality is that they need both votes and criticizing them doesn't help Democrats move forward with the bill.

Sen. Martin Heinrich (D-N.M.) declared: I cant respond to everything Joe Manchin says, thats not my job.

Zinger: Asked if he was aligned with Manchin on reconciliation, Sen. Jon Tester (D-Mont.) told Burgess and Marianne: Are you crazy? Are you trying to get me shot? Id never, ever want to be aligned with Joe Manchin. My wife would divorce me.

More on the Democrats dilemma with Manchin and Sinema from Burgess and Marianne: https://politi.co/2XrDpEd

ACCOUNTABILITY ON AFGHANISTAN Secretary of State Antony Blinken is gearing up for a second day of being hammered by lawmaker questions on the fall of Kabul and the American evacuation effort.

Blinken is a careful and calculated witness, offering lawmakers the deference they like along with a vigorous defense of the Biden administration. His style can be attributed to his time on Capitol Hill, serving as then-Sen. Joe Bidens staff when the Delawarean chaired the Senate Foreign Relations Committee.

Read up on Blinkens appearance before House Foreign Affairs Monday from Andrew Desiderio before this morning's hearing at Senate Foreign Relations.

GOOD MORNING! Welcome to Huddle, the play-by-play guide to all things Capitol Hill, on this Tuesday, September 14, where some folks are googling What is the Met Gala?

MONDAYS MOST CLICKED: House Democrats plan 26.5 percent corporate rate as part of multitrillion tax hike

SENATE FRIDAY? The Senate could hit the road as soon as today, cutting what was already slated to be a short week due to Yom Kippur starting Wednesday, even shorter. Theres a chance that after the Senate takes up to six votes Tuesday on nominees to the Department of Education and district judges, they could hit the road after a cloture vote on the nomination of Veronica Rossman to the tenth circuit.

Another signal: The Senate Armed Services Committee briefing on Afghanistan scheduled for Wednesday has been moved up to Tuesday. Could those be jet fumes already?

A message from AT&T:

Accessible, affordable broadband helps communities reach their American Dream. Thats why were making a $2 billion, 3-year commitment toward helping close the digital divide, so more low-income families have the ability to succeed. Learn more.

CONFIRMATION CALCULATION Centrists in the Senate Democratic caucus are calling on President Joe Biden to nominate a new director for the Bureau of Alcohol, Tobacco and Firearms after his initial pick, David Chipman, was yanked from consideration amid hesitation from some lawmakers.

It's a controversial position. The last ATF nominee confirmed by the Senate was B. Todd Jones back in 2013 and the agency hasn't had a confirmed director since 2015. With the 2022 midterms set to heat up in the coming months, Democrats want to see someone confirmed quickly who wont risk becoming a political liability for the party. More on the ATF nomination calculation from Marianne and Burgess here: https://politi.co/3htOBqK

THREADING THE GOP NEEDLE ON JAN 6 The rally planned for Saturday to support imprisoned pro-Trump rioters who stormed the Capitol on Jan. 6 is putting Republican leaders in a squeeze: They dont want to be seen as backing the insurrectionists, but theres a significant slice of the GOP base that justifies and supports the violent attack on the Captiol. Dont miss Olivias dive into these dynamics and more: https://politi.co/3hvVJTB

WHATEVER MAY HAPPEN Congressional leaders are expressing confidence in the Capitol Police ahead of this weekends rally after a briefing from security leaders Monday. But their comments also belied the dissatisfaction with how Jan. 6 played out.

House Speaker Nancy Pelosi (D-Calif.) said the the plans laid out by Capitol Police Chief Thomas Manger and House Sergeant-at-Arms William Walker seems much better. I dont have anything to compare it to because we werent briefed before.

"Much better prepared than before Jan. 6. I think they're ready for whatever might happen," Majority Leader Chuck Schumer (D-N.Y.) said after the briefing.

Manger said the fence around the Capitol will go up Friday and if everything goes well, it will come down very soon after.

PAGING: SENATE PAGES The civic-minded teens clad in blue polyester are BACK, baby! A sign that more and more normalcy is inching back to Capitol Hill. They were spotted Monday, eyes wide and mouths masked, getting acquainted with the maze of the Senate basement.

The last class of Senate pages had a flash in the pan experience: in the spotlight delivering messages, water, milk and more during the first impeachment trial of President Donald Trump and then sent home weeks later as a mysterious illness spread across the country.

Your Huddle host welcomes the return of youthful exuberance, the teen awkwardness and earnest interest in government! Heres to hoping the fall page class makes it through the semester (and that theyre all vaccinated.)

POLITICS AT THE FASHION FUNCTION Face it, Washington is not known for high fashion. But Reps. Alexandria Ocasio-Cortez (D-N.Y.) and Carolyn Maloney (D-N.Y.) made a splash Monday night at the return of the Met Gala, leaning into the theme of In America: A Lexicon of American Fashion.

Ocasio-Cortez took a central idea of the Democrats reconciliation package to the Met, sporting a white dress with Tax the Rich splashed down the back. "We really started having a conversation about what it means to be working-class women of color at the Met," said Ocasio-Cortez in a red carpet interview with Vogue. While the Met is known for its spectacle, we should have a conversation about it."

Maloney wore a dress calling for the ratification of the Equal Rights Amendment, which would amend the Constitution to prohibit discrimination based on sex. Maloney's gown included several cascading sashes reading "Equal rights for women," and she carried a green purse that read "ERA YES," mimicking iconic signage held by amendment supporters. The white, green and purple were a nod to the suffragist movement

In 2019 Maloney stunned the high-glam event by showing up in a New York Fire Department jacket, an effort to promote a bill that would for decades authorize the Sept. 11th Victim Compensation Fund.

QUICK LINKS

Shes One Of Congresss Leading Progressives Just Not In Her Own Office, Staffers Say, from BuzzFeed

Obamas, Bushes and Clintons teaming up in effort to aid Afghan refugees, from CNN

The Expanded Child Tax Credit Was a Godsend to Struggling Families. Will Democrats Save It? from Grace Segers at The New Republic

TRANSITIONS

Librarian of Congress Carla Hayden has appointed Judith Conklin as the chief information officer of the Library of Congress and John Rutledge the deputy chief information officer.

Sarah Shapiro was promoted to legislative director for Rep. Eric Swalwell (D-Calif.). She most recently was Swalwells policy advisor.

Jeremy Crane was promoted to be press secretary for Rep. Matt Rosendale (R-Mont.). He most recently was deputy press secretary for Rosendale.

Alejandra (Allie) Rodriguez is now scheduler and legislative correspondent for Rep. Mara Elvira Salazar (R-Fla.). She most recently was a legislative correspondent for Salazar.

Adam Farris has joined Sen. Tim Scotts (R-S.C.) legislative team handling tax and trade. He most recently was legislative director for Rep. Byron Donalds (R-Fla.).

TODAY IN CONGRESS

The House convenes at 11 a.m. for a pro forma session.

The Senate convenes at 10 a.m. with votes scheduled for 11:30 a.m., 2:20 p.m. and 5:30 p.m.

AROUND THE HILL

10 a.m. Secretary of State Antony Blinken testifies before the Senate Foreign Relations Committee.

1 p.m. Civil rights leaders including Rev. Al Sharpton, Martin Luther King III and others, including members of congress, hold a rally calling on the Senate to act on voting rights legislation.

2 p.m. Senate Democrats and Republicans hold their separate post-policy lunch press conferences.

TRIVIA

MONDAYS WINNER: Casey Burgat correctly answered that Madisons failed 12th amendment limited the size of congressional districts so that each one could contain no more than 50,000 citizens.

TODAYS QUESTION from Casey: At 607, Abraham Lincoln has more U.S. public schools named after him than any other president. To fill out the Mount Rushmore of this category, name the next three presidents with the most schools named in their honor.

The first person to correctly guess gets a mention in the next edition of Huddle. Send your answers to [emailprotected]

GET HUDDLE emailed to your phone each morning.

Follow Katherine on Twitter @ktullymcmanus

A message from AT&T:

Brooke Drydens daughter was diagnosed with learning disabilities at an early age. She requires an individualized education plan and weekly therapy with speech specialists. However, rural Colorado does not have the kind of specialists she needs. With the help of accessible and affordable broadband, Brooke is able to ensure that she receives regular virtual therapy and never falls behind. Brookes dream is to see her daughter not just survive but thrive in the world. Thats why AT&T is dedicated to helping close the digital divide with a $2 billion, 3-year commitment, so more low-income families like Brooke's can achieve their American Dream. Learn more.

The rest is here:
Dems bite their tongues on Manchin and Sinema - Politico

Blood Brothers: Malcolm X & Muhammad Ali (Netflix film) This documentary shines light on the friendship of the iconic twin-figures of Malcolm X and Muhammad Ali, whose extraordinary friendship (and falling out) reverberated throughout the 20th century. Kenya Barris and director Marcus A. Clarke celebrate the legendary bond in question with the help of family members and luminaries including Cornel West and Al Sharpton. In the end, their positions within the Nation of Islams leadership sees ups and downs and, overall, a legendary retelling of what really went down.

Kin: (AMC+ series) Charlie Cox stars in this new Irish series about a tight-knit crime family thats drawn into wartime mode with a mighty drug kingpin. The family soon discovers that this is an unwinnable war, yet as the losses begin to mount, it becomes clear that the cartel is at one distinct disadvantage: theyre not bound by unbreakable blood bonds. The cartel does, however, have a host of exotic pets and some snazzy costumes and celebrity status, so this is a heck of a story.

Frogger: Season 1 (Peacock series) The classic, squish-filled arcade game gets the adaptation treatment from Holey Moley producers Eureka Productions. Damon Wayans Jr. hosts as contestants attempt to conquer a series of lily pads while attempting to win the cash prizes that are, surely, more valuable to them than the moving vehicles or flooded streets that they attempt to conquer. Obstacles will include Frogs in Space and Frog Skull Island, so you cant resist watching at least once.

It Couldnt Happen Here: Season 1 (Sundance TV and AMC+ series) Hilarie Burton Morgan hosts this look at stories that have taken a back seat to the more sensational true-crime accounts out there. In doing so, she helps to examine local stories that tear apart the very fabric of the communities in which they surface, with a glimpse of the unique challenges of each afflicted community.

What We Do in the Shadows (FX, 10:00 & 10:30pm) Well, well, well. Guillermo turned out to be a vampire killer, which sure as heck came as a surprise to Nandor, Nadja, and Laszlo, and Colin. The four Staten Island roommates must figure out how to handle this conundrum, along with tackling the other challenges of this season. Those include dealing with wellness cults and gym culture, along with gargoyles, werewolves who play kickball, casinos, and more. In addition, the vamps also receive a higher level of powers while Nandor experiences an eternal-life crisis, which forces him to examine whether he should be a bachelor for eternity or embrace love. This week, an ancient vehicle and an old flame both see resurrection.

The Late Show With Stephen Colbert Sarah Paulson, Kacey Musgraves

The Other Two: Season 1 (HBO Max series) Lorne Michaels of SNL fame executive produces this series thats created, written, and also executive produced by Chris Kelly and Sarah Schneider (formerly co-head writers of SNL). The cast includes Drew Tarver, Helne Yorke, Case Walker, Ken Marino, and Molly Shannon, and the plot follows a showbiz family, in which a 14-year-old pop star decides that its time to officially retire. Meanwhile, the familys 53-year-old matriarch (Shannon) is enjoying ubiquity of her own, so The Other Two will do everything they can to shine as well.

Star Trek: Lower Decks: Season 2 (Paramount+ series) This animated series from Rick and Morty writer (and Solar Opposites creator) Mike McMahan takes things to the year 2380 (after the original Star Trek beginning in 2265), where the U.S.S. Cerritos arent the heroes that youre expecting. These are junior officers who are not pleased at their lack of power while confronting bizarre alien anomalies like enormous bugs and other such comedic-slanted creatures. This violent shows got a PG-13-like feel.

See more here:
What's On Tonight: 'Blood Brothers' & 'Frogger' Span Serious-To-Silly - UPROXX

The Transform Technology Summits start October 13th with Low-Code/No Code: Enabling Enterprise Agility. Register now!

Will deep learning really live up to its promise? We dont actually know. But if its going to, it will have to assimilate how classical computer science algorithms work. This is what DeepMind is working on, and its success is important to the eventual uptake of neural networks in wider commercial applications.

Founded in 2010 with the goal of creating AGI artificial general intelligence, a general purpose AI that truly mimics human intelligence DeepMind is on the forefront of AI research. The company is also backed by industry heavyweights like Elon Musk and Peter Thiel.

Acquired by Google in 2014, DeepMind has made headlines for projects such as AlphaGo, a program that beat the world champion at the game of Go in a five-game match, and AlphaFold, which found a solution to a 50-year-old grand challenge in biology.

Now DeepMind has set its sights on another grand challenge: bridging the worlds of deep learning and classical computer science to enable deep learning to do everything. If successful, this approach could revolutionize AI and software as we know them.

Petar Velikovi is a senior research scientist at DeepMind. His entry into computer science came through algorithmic reasoning and algorithmic thinking using classical algorithms. Since he started doing deep learning research, he has wanted to reconcile deep learning with the classical algorithms that initially got him excited about computer science.

Meanwhile, Charles Blundell is a research lead at DeepMind who is interested in getting neural networks to make much better use of the huge quantities of data theyre exposed to. Examples include getting a network to tell us what it doesnt know, to learn much more quickly, or to exceed expectations.

When Velikovi met Blundell at DeepMind, something new was born: a line of research that goes by the name of Neural Algorithmic Reasoning (NAR), after a position paper the duo recently published.

NAR traces the roots of the fields it touches upon and branches out to collaborations with other researchers. And unlike much pie-in-the-sky research, NAR has some early results and applications to show for itself.

Velikovi was in many ways the person who kickstarted the algorithmic reasoning direction in DeepMind. With his background in both classical algorithms and deep learning, he realized that there is a strong complementarity between the two of them. What one of these methods tends to do really well, the other one doesnt do that well, and vice versa.

Usually when you see these kinds of patterns, its a good indicator that if you can do anything to bring them a little bit closer together, then you could end up with an awesome way to fuse the best of both worlds, and make some really strong advances, Velikovi said.

When Velikovi joined DeepMind, Blundell said, their early conversations were a lot of fun because they have very similar backgrounds. They both share a background in theoretical computer science. Today, they both work a lot with machine learning, in which a fundamental question for a long time has been how to generalize how do you work beyond the data examples youve seen?

Algorithms are a really good example of something we all use every day, Blundell noted. In fact, he added, there arent many algorithms out there. If you look at standard computer science textbooks, theres maybe 50 or 60 algorithms that you learn as an undergraduate. And everything people use to connect over the internet, for example, is using just a subset of those.

Theres this very nice basis for very rich computation that we already know about, but its completely different from the things were learning. So when Petar and I started talking about this, we saw clearly theres a nice fusion that we can make here between these two fields that has actually been unexplored so far, Blundell said.

The key thesis of NAR research is that algorithms possess fundamentally different qualities to deep learning methods. And this suggests that if deep learning methods were better able to mimic algorithms, then generalization of the sort seen with algorithms would become possible with deep learning.

To approach the topic for this article, we asked Blundell and Velikovi to lay out the defining properties of classical computer science algorithms compared to deep learning models. Figuring out the ways in which algorithms and deep learning models are different is a good start if the goal is to reconcile them.

For starters, Blundell said, algorithms in most cases dont change. Algorithms are comprised of a fixed set of rules that are executed on some input, and usually good algorithms have well-known properties. For any kind of input the algorithm gets, it gives a sensible output, in a reasonable amount of time. You can usually change the size of the input and the algorithm keeps working.

The other thing you can do with algorithms is you can plug them together. The reason algorithms can be strung together is because of this guarantee they have: Given some kind of input, they only produce a certain kind of output. And that means that we can connect algorithms, feeding their output into other algorithms input and building a whole stack.

People have been looking at running algorithms in deep learning for a while, and its always been quite difficult, Blundell said. As trying out simple tasks is a good way to debug things, Blundell referred to a trivial example: the input copy task. An algorithm whose task is to copy, where its output is just a copy of its input.

It turns out that this is harder than expected for deep learning. You can learn to do this up to a certain length, but if you increase the length of the input past that point, things start breaking down. If you train a network on the numbers 1-10 and test it on the numbers 1-1,000, many networks will not generalize.

Blundell explained, They wont have learned the core idea, which is you just need to copy the input to the output. And as you make the process more complicated, as you can imagine, it gets worse. So if you think about sorting through various graph algorithms, actually the generalization is far worse if you just train a network to simulate an algorithm in a very naive fashion.

Fortunately, its not all bad news.

[T]heres something very nice about algorithms, which is that theyre basically simulations. You can generate a lot of data, and that makes them very amenable to being learned by deep neural networks, he said. But it requires us to think from the deep learning side. What changes do we need to make there so that these algorithms can be well represented and actually learned in a robust fashion?

Of course, answering that question is far from simple.

When using deep learning, usually there isnt a very strong guarantee on what the output is going to be. So you might say that the output is a number between zero and one, and you can guarantee that, but you couldnt guarantee something more structural, Blundell explained. For example, you cant guarantee that if you show a neural network a picture of a cat and then you take a different picture of a cat, it will definitely be classified as a cat.

With algorithms, you could develop guarantees that this wouldnt happen. This is partly because the kind of problems algorithms are applied to are more amenable to these kinds of guarantees. So if a problem is amenable to these guarantees, then maybe we can bring across into the deep neural networks classical algorithmic tasks that allow these kinds of guarantees for the neural networks.

Those guarantees usually concern generalizations: the size of the inputs, the kinds of inputs you have, and their outcomes that generalize over types. For example, if you have a sorting algorithm, you can sort a list of numbers, but you could also sort anything you can define an ordering for, such as letters and words. However, thats not the kind of thing we see at the moment with deep neural networks.

Another difference, which Velikovi noted, is that algorithmic computation can usually be expressed as pseudocode that explains how you go from your inputs to your outputs. This makes algorithms trivially interpretable. And because they operate over these abstractified inputs that conform to some preconditions and post-conditions, its much easier to reason theoretically about them.

That also makes it much easier to find connections between different problems that you might not see otherwise, Velikovi added. He cited the example of MaxFlow and MinCut as two problems that are seemingly quite different, but where the solution of one is necessarily the solution to the other. Thats not obvious unless you study it from a very abstract lens.

Theres a lot of benefits to this kind of elegance and constraints, but its also the potential shortcoming of algorithms, Velikovi said. Thats because if you want to make your inputs conform to these stringent preconditions, what this means is that if data that comes from the real world is even a tiny bit perturbed and doesnt conform to the preconditions, Im going to lose a lot of information before I can massage it into the algorithm.

He said that obviously makes the classical algorithm method suboptimal, because even if the algorithm gives you a perfect solution, it might give you a perfect solution in an environment that doesnt make sense. Therefore, the solutions are not going to be something you can use. On the other hand, he explained, deep learning is designed to rapidly ingest lots of raw data at scale and pick up interesting rules in the raw data, without any real strong constraints.

This makes it remarkably powerful in noisy scenarios: You can perturb your inputs and your neural network will still be reasonably applicable. For classical algorithms, that may not be the case. And thats also another reason why we might want to find this awesome middle ground where we might be able to guarantee something about our data, but not require that data to be constrained to, say, tiny scalars when the complexity of the real world might be much larger, Velikovi said.

Another point to consider is where algorithms come from. Usually what happens is you find very clever theoretical scientists, you explain your problem, and they think really hard about it, Blundell said. Then the experts go away and map the problem onto a more abstract version that drives an algorithm.The experts then present their algorithm for this class of problems, which they promise will execute in a specified amount of time and provide the right answer. However, because the mapping from the real-world problem to the abstract space on which the algorithm is derived isnt always exact, Blundell said, it requires a bit of an inductive leap.

With machine learning, its the opposite, as ML just looks at the data. It doesnt really map onto some abstract space, but it does solve the problem based on what you tell it.

What Blundell and Velikovi are trying to do is get somewhere in between those two extremes, where you have something thats a bit more structured but still fits the data, and doesnt necessarily require a human in the loop. That way you dont need to think so hard as a computer scientist. This approach is valuable because often real-world problems are not exactly mapped onto the problems that we have algorithms for and even for the things we do have algorithms for, we have to abstract problems. Another challenge is how to come up with new algorithms that significantly outperform existing algorithms that have the same sort of guarantees.

When humans sit down to write a program, its very easy to get something thats really slow for example, that has exponential execution time, Blundell noted. Neural networks are the opposite. As he put it, theyre extremely lazy, which is a very desirable property for coming up with new algorithms.

There are people who have looked at networks that can adapt their demands and computation time. In deep learning, how one designs the network architecture has a huge impact on how well it works. Theres a strong connection between how much processing you do and how much computation time is spent and what kind of architecture you come up with theyre intimately linked, Blundell said.

Velikovi noted that one thing people sometimes do when solving natural problems with algorithms is try to push them into a framework theyve come up with that is nice and abstract. As a result, they may make the problem more complex than it needs to be.

The traveling [salesperson], for example, is an NP complete problem, and we dont know of any polynomial time algorithm for it. However, there exists a prediction thats 100% correct for the traveling [salesperson], for all the towns in Sweden, all the towns in Germany, all the towns in the USA. And thats because geographically occurring data actually has nicer properties than any possible graph you could feed into traveling [salesperson], Velikovi said.

Before delving into NAR specifics, we felt a naive question was in order: Why deep learning? Why go for a generalization framework specifically applied to deep learning algorithms and not just any machine learning algorithm?

The DeepMind duo wants to design solutions that operate over the true raw complexity of the real world. So far, the best solution for processing large amounts of naturally occurring data at scale is deep neural networks, Velikovi emphasized.

Blundell noted that neural networks have much richer representations of the data than classical algorithms do. Even inside a large model class thats very rich and complicated, we find that we need to push the boundaries even further than that to be able to execute algorithms reliably. Its a sort of empirical science that were looking at. And I just dont think that as you get richer and richer decision trees, they can start to do some of this process, he said.

Blundell then elaborated on the limits of decision trees.

We know that decision trees are basically a trick: If this, then that. Whats missing from that is recursion, or iteration, the ability to loop over things multiple times. In neural networks, for a long time people have understood that theres a relationship between iteration, recursion, and the current neural networks. In graph neural networks, the same sort of processing arises again; the message passing you see there is again something very natural, he said.

Ultimately, Blundell is excited about the potential to go further.

If you think about object-oriented programming, where you send messages between classes of objects, you can see its exactly analogous, and you can build very complicated interaction diagrams and those can then be mapped into graph neural networks. So its from the internal structure that you get a richness that seems might be powerful enough to learn algorithms you wouldnt necessarily get with more traditional machine learning methods, Blundell explained.

Go here to read the rest:
DeepMind aims to marry deep learning and classic algorithms - VentureBeat

Seventeen years after the Bonn Agreement under which a new interim administration was established, .

. Unemployment rates are high. The countrys rapid population growth places pressure on service delivery and .

Declines in grant assistance accompanying the drawdown of international security forces has weakened demand and led to a broad and sustained economic slowdown.

While much progress has been made, institutions do not adequately mediate competition and conflict over resources, protect property rights, or keep citizens safe.

that are difficult to generate.

In this context, how can economic development be achieved in Afghanistan?

The Afghanistan to 2030 report highlights a set of priorities for economic development in Afghanistan, taking ongoing fragility as a given.

The report answers the following questions:

The report finds that .

This would require policy measures to support households and businesses deal with the risks of insecurity. It would also require a balanced growth strategy, involving increased public spending on human capital, improved agricultural productivity, and the mobilization of new investment in the extractives sector.

The report draws on several background papers that are available below.

See the original post here:
Afghanistan to 2030: Priorities for Economic Development ...