Alphabet Inc’s shareholders, including top executives, voted down several proposals on Wednesday, defeating campaigns to tie pay to diversity goals and to get the Google parent to provide more data about efforts to moderate user-generated content.
arXiv, Computer Science > Computers and Society; Cristina Kadar and Irena Pletikosa
from
Traditional crime prediction models based on census data are limited, as they fail to capture the complexity and dynamics of human activity. With the rise of ubiquitous computing, there is the opportunity to improve such models with data that make for better proxies of human presence in cities. In this paper, we leverage large human mobility data to craft an extensive set of features for crime prediction, as informed by theories in criminology and urban studies. We employ averaging and boosting ensemble techniques from machine learning, to investigate their power in predicting yearly counts for different types of crimes occurring in New York City at census tract level. Our study shows that spatial and spatio-temporal features derived from Foursquare venues and checkins, subway rides, and taxi rides, improve the baseline models relying on census and POI data. The proposed models achieve absolute R^2 metrics of up to 65% (on a geographical out-of-sample test set) and up to 89% (on a temporal out-of-sample test set). This proves that, next to the residential population of an area, the ambient population there is strongly predictive of the area’s crime levels. We deep-dive into the main crime categories, and find that the predictive gain of the human dynamics features varies across crime types: such features bring the biggest boost in case of grand larcenies, whereas assaults are already well predicted by the census features. Furthermore, we identify and discuss top predictive features for the main crime categories. These results offer valuable insights for those responsible for urban policy or law enforcement.
The Alderman Library was the reason Geeta Patel came to the University of Virginia.
Alderman is the university’s main library, and its open stacks were what attracted Patel, an associate professor whose work, in languages and in women and gender studies, relies heavily on archives, to Charlottesville.
But plans to renovate Alderman — and cut shelf space in the process — have Patel and others worried that the library’s strength will be diminished.
Designs for the renovation, which will come before a committee of the university’s Board of Visitors on Thursday, call for a 40 percent to 70 percent reduction in the library’s shelf capacity, depending on what type of shelving is used.
World-beating, people-centered companies have long been birthed and thrive in Toronto, including Four Seasons Hotels and Resorts and MAC Cosmetics. It’s a natural home for the kind of people who really start companies (rather than the people who start companies in the movies): midcareer men and women, who care not only about their work but also about their families. Toronto’s health and family-leave benefits are more generous than those found in the United States. That, coupled with its strong school system, does wonders in terms of providing opportunities by taking certain key stresses out of the equation.
It also acts as a giant magnet for the academically ambitious, not only from within Canada but from all over the world. Its slew of world-class universities, including the University of Toronto, York University, and, just to the west, the University of Waterloo and the University of Western Ontario, attract tens of thousands of smart students every year. As they graduate and look to settle down and start families, many, happy with the lifestyle Canada’s largest city affords, stay in Toronto. As a result, founders there can hire people they’d never be able to attract in Silicon Valley.
I started my remote internship with LC Labs expecting to build a Twitterbot to support the Library of Congress Baseball Americana engagement with MLB All-Star Week 2018. Around the same time, I was ramping up an unrelated school project to build a chatbot prototype that would connect visitors on Governors Island with stories about its historic district. My aim in that undertaking was to use the physical context of the island to seamlessly engage New Yorkers with interesting, localized digital content using an AI tool I hadn’t much explored up to that point. What I soon realized however was that if my aim was to connect Gov Island history with its context in space, I needed first to connect more meaningfully with the digital collections that I was channeling history from. I didn’t feel right about asserting that this ‘place-based’ approach to storytelling would be worth exploring without first establishing an appreciation for the digital collections I’d be serving up. It dawned on me that my internship with the Library could be the perfect opportunity to meet with and learn from the staff behind these collections, as well as other members of LC Labs who were leveraging the newly launched loc.gov JSON API to provide machine-readable access to the collections for a variety of apps and purposes.
As I pivoted with my primary objective for the internship, one of the first things I decided on with the chatbot prototype was to focus on just a select few primary sources as data sources for the stories I’d be crafting for the bot to tell. During an in-person visit early on in my internship, I met with key folks from Chronicling America (ChronAm), of the Serial and Government Publications Division and the Veterans History Project (VHP), of the American Folklife Center.
In meeting with Robin Butterhof of ChronAm, I picked up on intricacies about how historical newspapers are collected and stored that later became critical for the prototype as I figured out how to programmatically access, manipulate, and deliver newspaper content within the Facebook Messenger environment.
There is an urgent need for more effective treatments for major depressive disorder (MDD). Digital therapeutics, such as computerized cognitive–emotional training interventions, represent a promising new strategy for treating MDD. Here we report a replication of efficacy of a digital cognitive–emotional training intervention designed to enhance cognitive control for emotional information-processing. In a randomized, double-blind, controlled study design, 51 participants with MDD in a current major depressive episode were randomly assigned to participate in a digital cognitive–emotional training regimen (Emotional Faces Memory Task (EFMT); n = 28) involving 18 sessions over 6 weeks, or an active control condition (CT; n = 23) involving computerized working memory training. MDD symptoms were assessed weekly using a clinician-rated measure (Hamilton Depression Rating Scale; Ham-D); and neurocognition (working memory), at baseline and study outcome. Mixed-effects model for repeated measures (MMRM) analysis of all participants randomized revealed a significantly greater reduction in MDD symptom severity (Ham-D) from baseline to outcome in the EFMT group (8.65 points) compared to the CT group (4.77 points) (F(6,205) = 3.23, p = .005, d = 0.46). Ten of 28 EFMT participants achieved clinical response (≥50% reduction in symptoms) compared to 4 of 23 in CT. Both groups exhibited similar, small improvements in working memory. This replicated the preliminary efficacy of a digital cognitive–emotional training approach for the treatment of MDD. EFMT may be a feasible and effective intervention strategy for MDD, but future studies to elucidate its mechanism of action are warranted. This study is registered with Clinicaltrials.gov (NCT: 01934491).
In an important step forward in the consumerization of healthcare, this week Apple introduced an update to its Health app (iOS 11.3 beta) which allows individuals to access medical records on their iPhone. Based on the FHIR (Fast Healthcare Interoperability Resources) standard, the company’s new Health Records API allows patient data from over 500 hospitals and clinics, including Johns Hopkins Medicine, Cedars-Sinai, and Penn Medicine, to be available through the Health app interface.
The move gives patients greater access and agency with personal health information by allowing individuals to share data from their health record with third party applications. The announcement noted that data shared is transferred directly from HealthKit to third party applications without touching Apple’s servers. This represents a move towards breaking down the traditional silos of medical record systems between which interoperability and data sharing has been an ongoing challenge.
We recognize that such powerful technology raises equally powerful questions about its use. How AI is developed and used will have a significant impact on society for many years to come. As a leader in AI, we feel a deep responsibility to get this right. So today, we’re announcing seven principles to guide our work going forward. These are not theoretical concepts; they are concrete standards that will actively govern our research and product development and will impact our business decisions.
We acknowledge that this area is dynamic and evolving, and we will approach our work with humility, a commitment to internal and external engagement, and a willingness to adapt our approach as we learn over time.
UCLA Newsroom, Opinion + Voices, Judea Pearl and Dana Mackenzie
from
Computer programs have reached a bewildering point in their long and unsteady journey toward artificial intelligence. They outperform people at tasks we once felt to be uniquely human, such as playing poker or recognizing faces in a crowd. Meanwhile, self-driving cars using similar technology run into pedestrians and posts and we wonder whether they can ever be trustworthy.
Amid these rapid developments and nagging setbacks, one essential building block of human intelligence has eluded machines for decades: Understanding cause and effect.
Put simply, today’s machine-learning programs can’t tell whether a crowing rooster makes the sun rise, or the other way around. Whatever volumes of data a machine analyzes, it cannot understand what a human gets intuitively. From the time we are infants, we organize our experiences into causes and effects. The questions “Why did this happen?” and “What if I had acted differently?” are at the core of the cognitive advances that made us human, and so far are missing from machines.
As satellite images become more ubiquitous, we should reflect on where they come from, how they are created, and the purpose for their use.
Extra Extra
The Bureau of Labor Statistics ran a survey about contract labor that it hasn’t run since 2005 and found that only about 10 percent of the working population is a full-time gig economy worker. This is far lower than the roughly 30 percent who earn any income from gig work. The tough part is that most of the people doing full-time gig work wish they had more traditional W2-issuing jobs.
StatsBomb, a global soccer data analysis and strategy consultancy, announced that it will give away the proprietary data it collects from women’s soccer matches. The firm hopes to increase attention on the women’s game, and to boost the number of working women analysts throughout the industry.
For most women, one of the most stressful parts of giving birth is not knowing when it’s going to happen. Roughly 15 million pregnant women face life-threatening spontaneous preterm birth every year. And doctors don’t really understand why some pregnancies — nearly 10 percent of all U.S. births — end suddenly, weeks or even months before they should.
Now, a pilot study from researchers at Stanford University suggests that it may soon be possible to use a blood test to improve predictions of both healthy and too-early due dates. Such predictions could also help better explain why some births begin in crisis.
“It’s really hard to understate the potential of what these folks are proposing,” said Dr. Thomas McElrath, a maternal-fetal medicine specialist at Brigham and Women’s Hospital and an associate professor at Harvard Medical School in Boston. “The potential for how that may feed into care and into research and into furthering not just maternal but neonatal outcomes is huge. We’re just probably beginning to get a sense of what that may involve.”
Two new papers on fees and investor behavior— one by Morningstar and NORC at the University of Chicago and the other by the Financial Conduct Authority of the United Kingdom—can help us understand how investors respond to investment fees. In both reports, the authors find that individual investors partially ignore the fees they are paying, with potentially serious consequences. By changing the way in which investments are presented and chosen, however, investors would be more likely to take fees into account. Let’s take a quick look at the findings from each paper and examine what they mean, especially in light of our research that finds that the average asset-weighted fees paid by investors in the United States are declining.
Chicago, IL U.S. Soccer will host an inaugural U.S. Soccer Hackathon on July 14-15 in Chicago at 1871. Deadline for registration applications is June 29.
Today Apple delivered a Health Records API for developers and researchers to create an ecosystem of apps that use health record data to better manage medications, nutrition plans, diagnosed diseases and more. The Health Records feature allows patients of more than 500 hospitals and clinics to access medical information from various institutions organized into one view on their iPhone. For the first time, consumers will be able to share medical records from multiple hospitals with their favorite trusted apps, helping them improve their overall health.
A.I. bots in gaming are usually built by hand-coding a bunch of rules that impart game-intelligence. For the most part, this approach does a fairly good job of making the bot imitate human-like behavior. However, for most games it is still easy to tell apart a bot from an actual human playing. If we want to make these bots behave more human-like, would it help to not build them using hand-coded rules? What if we simply let the bot figure out the game by making it learn from looking at how humans play?
Exploring this would require a game where it is possible to collect such data of humans playing the game ahead of developing the game itself. FIFA is one such game that let me explore this. Being able to play the game and record my in-game actions and decisions allowed me to train an end-to-end Deep Learning based bot without having to hard-code a single rule of the game.