Thanks to artificial intelligence and new machine-learning algorithms, facial recognition has moved beyond security applications, popularized by Hollywood as an easy way to spot bad guys. New facial recognition methods can not only identify people, but they can also determine their age, gender, ethnicity, emotional state, and whether they are “live” and in person.
One company leading these developments is Ever AI, which today announced enhancements to its facial recognition product suite. Among the new features is “liveness detection”, which can determine if a person’s face is “live” as opposed to a photo.
Ever AI said attackers have attempted to bypass facial recognition authentication methods by using photos of someone’s face printed on paper, or a mobile device screen image, to try to gain access to a computer or mobile device.
Applied Geography; Hung Chak Ho, Anders Knudby, Guangqing Chi, Mehdi Aminipouri, Derrick Yuk-Fo Lai
from
Excess mortality can be caused by extreme hot weather events, which are increasing in severity and frequency in Canada due to climate change. Individual and social vulnerability factors influence the mortality risk associated with a given heat exposure. We constructed heat vulnerability indices using census data from 2006 to 2011 in Canada, developed a novel design to compare spatiotemporal changes of heat vulnerability, and identified locations that may be increasingly vulnerable to heat. The results suggest that 1) urban areas in Canada are particularly vulnerable to heat, 2) suburban areas and satellite cities around major metropolitan areas show the greatest increases in vulnerability, and 3) heat vulnerability changes are driven primarily by changes in the density of older ages and infants. Our approach is applicable to heat vulnerability analyses in other countries.
Now the Health IT lead at Alphabet subsidiary Verily, Dr. Joshua Mandel was part of the team that created SMART Health IT, the open, standards-based platform that many health developers now use to build apps. But all that started when he got sick of some grunt work he was assigned as a medical student. Mandel told the story at Dev4Health at the HIMSS Innovation Center in Cleveland this week, during an onstage interview with Health 2.0 founder Matthew Holt.
“As a medical student, one of my tasks was to show up at the hospital at 5 o’ clock in the morning so I could copy down vital sign readings from the surgery board onto paper so we could have efficient rounds, knowing what are the most important cases,” he said. “This was about an hour-long task with very little educational value to it. It sucked.”
So Mandel spent what little free time he had building a system to automate the task.
Twitter is telling all users that they should change their passwords, now.
The company’s chief technology officer, Parag Agrawal, said in a blog post that it had recently found a bug that stored passwords, unmasked, in an internal log.
Facebook has problems. Fake news. Terrorism. Russian propaganda. And maybe soon regulation. The company’s solution: Turn them into artificial-intelligence problems. The strategy will require Facebook to make progress on some of the biggest challenges in computing.
During two congressional sessions last month, CEO Mark Zuckerberg referenced AI more than 30 times in explaining how the company would better police activity on its platform. The man tasked with delivering on those promises, CTO Mike Schroepfer, picked up that theme in a keynote and interview at Facebook’s annual developer conference Wednesday.
Schroepfer told thousands of developers and journalists that “AI is the best tool we have to keep our community safe at scale.” After the congressional hearings, critics accused Zuckerberg of invoking AI to mislead people into thinking the company’s challenges are simply technological. Schroepfer told WIRED Wednesday that the company had made mistakes. But he said that for Facebook—with more than 2 billion people on its service each month—AI is the only way to address them.
By September, companies will have the specifications they need to manufacture 5G-enabled base stations
University Data Science News
Ali Rahimi thinks machine learning research is looking too much like “alchemy” and Francois Chollet thinks it may be sliding into “cargo cult” territory, in which pet methods for optimization are adopted with no deep rigor. This is the best piece of sensible, critical discussion on the state of machine learning from the week. It concludes with a dissenting opinion from Yann LeCun, who is comfortable with the messiness.
What?? Washington University in St. Louis has decided to start charging fees for Institutional Review Board applications. Five hundred bucks for a modification and $1500 for a continuing review. This is particularly alarming in data science where plenty of researchers are confused about whether they have to work with IRB on data science proposals – the answer is usually ‘yes’ because data come from people and are applied back into human contexts (unless you are an astronomer). Washington University is not poor. It is a private university with an endowment of $7.5 billion putting it in the top 1% of university endowments. I am completely baffled by this course of action.
A new paper presented at CHI found that Amazon Mechanical Turk workers typically earn no more than $2 per hour. This is ethically gross. Please, think twice before relying on extremely low wage labor to run your scientific experiments. It’s exploitative. Please don’t send an email suggesting turkers should be happy with the pittance or choose to work elsewhere. Let’s put poor homo economicus, the rational actor to rest already.
One rational economic decision that I cannot argue with is the new slate of short executive education coursesMIT is offering in natural language processing, robotics, and machine learning. Prepare to shell out for these.
Pittsburgh is increasingly becoming a hub for automation, robotics, and data science. The Army Research Laboratory just established a research partnership with Carnegie Mellon University via its Open Campus Initiative, Facebook decided to open another Facebook AI Research (FAIR) group in Pittsburgh, and Pittsburgh is still on the list of potential Amazon HQ2 cities. (It’s true, most people think Amazon HQ2 is going to DC/Northern Virginia which I still believe and have been saying since before the list narrowed to 20 cities.)
JP Morgansnapped up one of Carnegie Mellon University’s top machine learning researchers – Manuela Veloso – who will take a leave of absence from the school.
The Mellon Foundation gave Northeastern University$200,000 to found a center for the study of Boston. Two-hundred thousand seems like a seed grant? A feasibility study? A carrel in the library where one person reads a lot, blogs, and posts pictures of their weekend walks along Mass Ave? Funders and wealthy Bostonians, please give these scholars more money. Studying cities in depth is a great idea, but scholars, like Mechanical Turk workers, need a living wage.
Cass Sunstein of Harvard did a preliminary study to see 1) how much current Facebook users would pay to continue to use the service and 2) how much it would take to get the same Facebook users to stop using the service. For an economist, this is basically asking the same question in two different ways, but we do not operate like economists. Sunstein found most people were willing to pay only $1 to keep using facebook but would have to be paid at least $50 to stop using it. He warns that this is all preliminary. I remind you that Zuck has always said facebook will always be free to use.
Princeton University researchers have decided to combat stupidity on the internet – specifically poorly secured IoT devices that send sensitive information around in the clear. Quoting from the write-up, “the width and depth of manufacturer incompetence on display can’t be understated”. Of course, there are many other sources of incompetence and stupidity on the internet so don’t let Princeton deter you from launching your own wisdom crusade.
Using teaching evaluations to assess teachers may be illegal because students consistently display (what we hope is subconscious) gender bias towards men.
Final word from the universities to the universities by Anna Lauren Hoffman writes,”If you have the temerity to insert your work into a political issue that, by and large, doesn’t immediately affect your life, you should also be prepared to accept the consequences — or, at the very least, answer a few hard questions.” No more of this — “I built an algorithm to predict who is a gang member” and then cowering behind “I’m just an engineer” responses — not when audience members ask you to convince them that you haven’t just created a minefield of destructive unintended consequences.
Google has kicked off a new investment program to back early-stage startups building capabilities and devices using the tech giant’s digital assistant, and a Seattle company is among the first batch of investments.
Pulse Labs, a Seattle startup that provides a testing platform for developers to connect with users, is one of four companies to participate in the Google Assistant investment program so far. As part of the program, Google joined a $2.5 million investment round for Pulse Labs in January. Ironically, that round also included participation from Google’s main voice assistant rival Amazon through the $100 million Alexa Fund and Bezos Expeditions (Jeff Bezos’ investment arm).
In addition to investment, the program includes advice from Google engineers, product managers, and design experts as well as promotion from Google’s marketing teams. Startups will gain access to the Google Cloud Platform and partnership programs to get first crack at new features and tools.
There are nearly 33,000 Alexa Skills available today. More than 6,800 have been added since the beginning of the year, and 1,500+ in the last month alone.
We continue to be big believers in voice — it’s the natural evolution of human computer interaction. Little kids already know how to interact with these devices. We see this with our own families and friends.
Princeton CITP, Freedom to Tinker blog, Julia Stoyanovich and Bill Howe
from
In this post, Julia Stoyanovich and Bill Howe discuss their recent technical progress on bringing the idea of Ranking Facts to life, placing the nutritional label metaphor in the broader context of the ongoing algorithmic accountability and transparency debate.
In studies with monkeys, Johns Hopkins researchers report that they have uncovered significant new details about how the cerebellum — the “learning machine” of the mammalian brain — makes predictions and learns from its mistakes, helping us execute complex motor actions such as accurately shooting a basketball into a net or focusing your eyes on an object across the room.
In a summary of the study published on April 16 in Nature Neuroscience, the investigators provide a better understanding of why degenerative diseases that affect the cerebellum cause people to lose control of their movements. Their results demonstrate that the cerebellum is organized in a very different way than current designs of artificial neural networks, which are currently used in machine learning and artificial intelligence.
As AI and machine learning systems advance, real-world applications are exposing limitations such as the inability to adapt to situations beyond the narrow tasks those systems were trained to perform.
Indeed, some observers note that AI researchers still do not understand the mechanisms by which automated systems learn.
As AI researchers grapple with these unknowns, the Defense Advanced Research Projects Agency (DARPA) launched a program last year aimed at advancing “lifelong machine learning.” This week, the research agency announced funding for three university research teams that will tackle various components of the adaptability problem.
Those efforts augment initial research at Columbia University focused on reproducing neural networks.
A prominent computer scientist and professor at Carnegie Mellon University who heads its machine learning department has accepted a position with J.P. Morgan Chase, where she will lead artificial intelligence research, the school confirmed Thursday.
Manuela Veloso is taking a leave of absence from the university, effective July 1, officials including Andrew Moore, dean of the school of computer science, said.
We’re proposing an AI safety technique which trains agents to debate topics with one another, using a human to judge who wins. We believe that this or a similar approach could eventually help us train AI systems to perform far more cognitively advanced tasks than humans are capable of, while remaining in line with human preferences. We’re going to outline this method together with preliminary proof-of-concept experiments and are also releasing a web interface so people can experiment with the technique.
The Brookings Institution, Roy Parizat and Heinz-Wilhelm Strubenhoff
from
Imagine a poor farmer in Ghana that has been growing beans to sell into the local market. After providing for her family, she doesn’t have enough money left over to purchase high quality seeds or fertilizers, even though such inputs would increase the amount of beans she will grow. She tried borrowing from the local microfinance bank, but since she had never saved with them before, they are not willing to grant her a loan. Suddenly she receives a text message asking if she’d like to apply for a loan to pay for inputs. She clicks “yes,” and receives an automated message saying she has received the loan. The loan automatically goes into her mobile money account, and she visits the supply store to purchase seeds and fertilizers. Next harvest she produces 30 percent more beans than usual. After selling her crops she visits the local mobile money agent to repay the loan and interest. As a result, she is left with more money than normal, so she joins the local microfinance bank to open a savings account with the surplus funds. Now, she is an individual with a bank account opening further opportunities for her and her family.
The story above is not yet happening but is a distinct possibility and illustrates how technology can transform a farmer into a profitable borrower. While cellphone-based lending is starting to happen in Africa, with products such as MShwari in Kenya, this type of lending so far has been mainly for small, short-term loans for urban customers.
Atlanta, GA June 4. “AbGradCon (Astrobiology Graduate Conference) provides a unique setting for astrobiology-inclined graduate students and early-career researchers to come together to share their research, collaborate, and network.” [application required]
“The Mount Sinai Department of Rehabilitation Medicine is proud to announce a partnership with MIT Hacking Medicine to bring you a unique opportunity.” … “Up to $500,000 in in-kind prizes through a Shark Tank style pitch competition for startups looking to win a fully-funded Clinical Trial of their product at the Mount Sinai Department of Rehabilitation in exchange for 0.5-5% equity depending on the size and scope of the trial.” Deadline for submissions is May 12.
“The MLPerf effort aims to build a common set of benchmarks that enables the machine learning (ML) field to measure system performance for both training and inference from mobile devices to cloud services. We believe that a widely accepted benchmark suite will benefit the entire community, including researchers, developers, builders of machine learning frameworks, cloud service providers, hardware manufacturers, application providers, and end users.” Deadline for submissions is July 31.
arXiv, Computer Science > Human-Computer Interaction; Hao Fang, Hao Cheng, Maarten Sap, Elizabeth Clark, Ari Holtzman, Yejin Choi, Noah A. Smith, Mari Ostendorf
from
“We present Sounding Board, a social chatbot that won the 2017 Amazon Alexa Prize. The system architecture consists of several components including spoken language processing, dialogue management, language generation, and content management, with emphasis on user-centric and content-driven design. We also share insights gained from large-scale online logs based on 160,000 conversations with real-world users.”
“We would like to give you a preview of the roadmap for PyTorch 1.0 , the next release of PyTorch. Over the last year, we’ve had 0.2, 0.3 and 0.4 transform PyTorch from a [Torch+Chainer]-like interface into something cleaner, adding double-backwards, numpy-like functions, advanced indexing and removing Variable boilerplate. At this time, we’re confident that the API is in a reasonable and stable state to confidently release a 1.0.”
We are delighted to announce a significant expansion of our Swift.org continuous integration testing system. Members of the Swift community have been hard at work to support Swift on a number of new platforms, and we have extended the Swift CI system to support community-hosted nodes for testing additional platforms.