Data Science newsletter – July 12, 2021

Deep neural networks will move past their shortcomings without help from symbolic artificial intelligence, three pioneers of deep learning argue in a paper published in the July issue of the Communications of the ACM journal.

In their paper, Yoshua Bengio, Geoffrey Hinton, and Yann LeCun, recipients of the 2018 Turing Award, explain the current challenges of deep learning and how it differs from learning in humans and animals. They also explore recent advances in the field that might provide blueprints for the future directions for research in deep learning.

Titled “Deep Learning for AI,” the paper envisions a future in which deep learning models can learn with little or no help from humans, are flexible to changes in their environment, and can solve a wide range of reflexive and cognitive problems.

Willis Towers Watson and The Nature Conservancy (TNC) have published a report that demonstrates how ecological forest management – which reduces the risk of severe wildfires – can be combined with insurance and significantly cut insurance costs.

Researchers at TNC and Willis Towers Watson were able to quantify insurance premium savings by modeling the impact on insurance of controlled burning and the thinning of overgrown forests. Those insurance savings could, in turn, be used to fund further investments in sustainable forest management.

According to the report, the practice known as ecological forestry – which includes prescribed burns and the removal of smaller trees and other vegetation in overgrown forests – leads to a 41% decrease in total insurance premiums for homes and a range of reductions for commercial property, while also reducing the likelihood of extreme wildfires.

Teck Resources Ltd and Sibanye Stillwater Ltd are leading the way for artificial intelligence investment among top mining companies according to our analysis of a range of GlobalData data.

Artificial intelligence has become one of the key themes in the mining sector of late, with companies hiring for increasingly more roles, making more deals, registering more patents and mentioning it more often in company filings.

These themes, of which artificial intelligence is one, are best thought of as “any issue that keeps a CEO awake at night”, and by tracking and combining them, it becomes possible to ascertain which companies are leading the way on specific issues and which are dragging their heels.

In what has become a growing battle between big tech companies and the people who use their products, iconic Apple co-founder Steve Wozniak recently spoke out in support of the right to repair movement.

In a recent Cameo video made in response to another major right to repair advocate Louis Rossman, Woz said his busy schedule had previously prevented him from getting deeply involved with the topic, he now felt it was time to speak out on something that has “really affected me emotionally.” After doing his own research on the issue, Woz said he “totally supports” the right to repair movement and that the people behind the movement are “doing the right thing.”

President Biden is preparing to issue an executive order on the right to repair, White House press secretary Jen Psaki said Tuesday. This would be the first time a president has weighed in on the ability for consumers to fix their own things; it would also be the first time that a president has taken concrete steps on the issue.

At the White House press briefing today, Psaki said that the U.S. Department of Agriculture and the Federal Trade Commission, at the direction of Biden, is working on new rules to increase competition in the farming industry, allow for greater competition between small farmers and big businesses, and “give farmers the right to repair their own equipment how they like.” Psaki said this would be part of an executive order that the Biden administration will sign that is focused more broadly on helping farmers. Bloomberg reported the news earlier Tuesday.

The UC Board of Regents is poised to approve the largest gift ever made to UC Berkeley — a $300 million, 14-story, 772-bed dorm across from campus that will not only provide transfer students a place to sleep but will offer a “transformative” experience that pays attention to their emotional, mental and physical needs.

The Helen Diller Anchor House project, which the Regents will vote on at their July 21-22 meeting, will offer luxurious living conditions never before seen on campus — and won’t cost UC Berkeley a penny to build. In addition, net revenues spun out from rents will pay for about $2.5 million in scholarships each year for 100 first-generation, lower-income students.

The catch? Building the project will involve the destruction of two historic structures as well as a 112-year-old rent-controlled building at 1921 Walnut St. About six people, some of whom have lived there for a decade or more, will lose their homes.

The data does not lie: in successive editions of the annual Times Higher Education World University Rankings, Western nations have been losing ground while the East rises.

In 2016, just two universities from mainland China made the world top 200. Today the number is seven – led by Tsinghua University in Beijing, which made history by entering the world top 20 for the first time in 2021.

In the same period, Hong Kong has increased its representation from three to five institutions and South Korea from four to seven. Singapore is home to one of the world’s fastest-rising universities, Nanyang Technological University, which now sits in the world top 50, having rather symbolically lured one of the US’ most distinguished scholars, former US National Science Foundation director Subra Suresh, to lead its next phase of development.

Engineering projects are notoriously hard to complete on-time, with project delays often theorised to propagate across interdependent activities. Here, we use a novel dataset consisting of activity networks from 14 diverse, large-scale engineering projects to uncover network properties that impact timely project completion. We provide empirical evidence of perturbation cascades, where perturbations in the delivery of a single activity can impact the delivery of up to 4 activities downstream, leading to large perturbation cascades. We further show that perturbation clustering significantly affects project overall delays. Finally, we find that poorly performing projects have their highest perturbations in high reach nodes, which can lead to largest cascades, while well performing projects have perturbations in low reach nodes, resulting in localised cascades. Altogether, these findings pave the way for a network-science framework that can materially enhance the delivery of large-scale engineering projects.

Our new analysis (https://pewrsr.ch/3hHDnhy) finds that a majority of congregations shared or live-streamed at least one sermon discussing the 2020 election and the COVID-19 pandemic, while about half shared at least one discussing structural racism in America

Teaming up with industry titans including Samsung, NVIDIA, and more, researchers at The University of Texas at Austin are launching 6G@UT, a new research center to lay the groundwork for 6G, the next generation of wireless technology.

5G is just emerging as the dominant cellular technology after years of research and innovation that includes important contributions by UT Austin wireless researchers and alumni. With new technologies on the horizon such as self-driving cars and air taxis, holographic video conferencing, ubiquitous robotics and immersive augmented reality, UT Austin is cementing its leadership in wireless innovation with this major research effort on 6G.

“The advances in both wireless communications and machine learning over the past decade have been incredible, but separate,” said 6G@UT Director Jeffrey Andrews, a professor in UT Austin’s Department of Electrical and Computer Engineering. “Coupled with vast new sensing and localization abilities, 6G will be defined by an unprecedented native intelligence, which will transform the ability of the network to provide incredible services.”

Stevens Institute of Technology is one of two universities in the country that have been awarded the first-ever National Science Foundation grant to create an industry-university cooperative research center devoted specifically to financial technology and science, the school announced last week.

Officials from Stevens will team with officials from Rensselaer Polytechnic Institute on the Center for Research toward Advancing Financial Technologies, or CRAFT, as it will be known.

CRAFT, which will be led by Steve Yang and George Calhoun of the School of Business at Stevens and Aparna Gupta of the Lally School of Management at Rensselaer Polytechnic Institute in Troy, New York, underscores the nation’s strategic investment in managing risks and unintended consequences of the emerging — and yet unknown — challenges facing the high-tech financial services industry.

Microscopists have long sought a way to produce high-quality, deep-tissue imaging of living subjects in a timely fashion. To date, they have had to choose between image quality or speed when looking into the inner workings of complex biological systems.

A better imaging system would have a powerful impact on researchers in biology and in neuroscience, experts say. Now Dushan N. Wadduwage, a John Harvard Distinguished Science Fellow in imaging at the FAS Center of Advanced Imaging, along with a team from MIT, has detailed a new technique that would make that possible in a report in Science Advances.

In the paper, the team presents a new process that uses computational imaging to get high-resolution images 100 to 1,000 times faster than other state-of-the-art technologies using complex algorithms and machine learning. The method can shorten a process that often takes months into a matter of days.

Homes and offices are only as solid as the ground beneath them. When that solid ground turns to liquid — as sometimes happens during earthquakes — it can topple buildings and bridges. The phenomenon is known as liquefaction, and it was a major feature of the 2011 earthquake in Christchurch, New Zealand, a magnitude 6.3 quake that killed 185 people and destroyed thousands of homes.

An upside of the Christchurch quake was that it was one of the most well-documented in history. Because New Zealand is seismically active, the city was instrumented with numerous sensors for monitoring earthquakes. Post-event reconnaissance provided a wealth of additional data on how the soil responded across the city.

“It’s an enormous amount of data for our field,” said researcher Maria Giovanna Durante. “We said, ‘If we have thousands of data points, maybe we can find a trend.'”

The University of Illinois Urbana-Champaign has launched a new AI in Medicine Certificate program through an interdisciplinary partnership between the department of bioengineering at The Grainger College of Engineering, the Carle Illinois College of Medicine and the University of Illinois College of Veterinary Medicine. This self-paced online certificate program will equip healthcare professionals with a conceptual understanding of AI and its applications through real-world medical case studies using machine learning models. The cost for this program is $750 and Continuing Medical Education (CME) credits are available upon completion issued by the Carle health system and the University of Illinois College of Veterinary Medicine.

Large scale studies of underwater noise during rain are important for assessing the ocean environment and enabling remote sensing of rain rates over the open ocean. In this study, approximately 3.5 yrs of acoustical and meteorological data recorded at the northeast Pacific continental margin are evaluated. The acoustic data are recorded at a sampling rate of 64 kHz and depths of 81 and 581 m at the continental shelf and slope, respectively. Rain rates and wind speeds are provided by surface buoys located in the vicinity of each hydrophone. Average power spectra have been computed for different rain rates and wind speeds, and linear and nonlinear regression have been performed. The main findings are (1) the linear regression slopes highly depends on the frequency range, rain rate, wind speed, and measurement depth; (2) noise levels during rain between 200 Hz and 10 kHz significantly increase with increasing wind speed; and (3) the highest correlation between the spectral level and rain rate occurs at 13 kHz, thus, coinciding with the spectral peak due to small raindrops. The results of this study indicate that previously proposed algorithms for estimating rain rates from acoustic data are not universally applicable but rather have to be adapted for different locations.