Data Science newsletter – May 21, 2021

A neural interface has been developed that could enable people with paralysis to type faster than they could using other technologies, by directly translating attempts at handwriting into text.

The academic integrity industry first started taking off in 2017 amid an uptick in distance learning and has continued to grow in the years since. The range of services offered by the sector include software that locks down browsers, programs that analyze papers for plagiarism and monitoring services that either make it easier for a human to watch a student take a test without physically being in the room or use artificial intelligence to keep an eye on students.

Data from 1,751 universities collected by ListedTech, a market research firm that tracks the use of education technology indicates that roughly 47% of these schools use one of these programs, nearly 25% use two and close to 15% use three or more.

It’s easy to see why instructors reached for these services in the early months of the pandemic. Faculty were asked to very quickly transform their in-person courses to online classes. Changing tests that are normally monitored in the classroom to be cheat-proof when taken from a student’s bedroom can be a heavy lift.

The new study draws on data from a global network of more than 500 Biogeochemical–Argo floats to analyze the locations, types, and drivers of deep chlorophyll maxima around the world. The free-drifting floats contain a variety of sensors, including chlorophyll sensors, and they return data from varying ocean depths worldwide.

Analysis of the float data reveals that the dynamics of deep chlorophyll maxima are dependent on location and season. For instance, deep chlorophyll maxima are seen permanently or frequently in tropical and subtropical regions. Meanwhile, at high latitudes, they are less frequent and occur mostly in the summer in the form of deep biomass maxima.

In low-nutrient (oligotrophic) ocean regions, the analysis detected deep photoacclimation maxima that are permanent features but that are sometimes replaced by deep biomass maxima in the summer. Waters just north or south of equatorial regions appear to host permanent deep biomass maxima, thanks to abundant light and nutrients.

Researchers at UC San Francisco have observed a new feature of neural activity in the hippocampus – the brain’s memory hub – that may explain how this vital brain region combines a diverse range of inputs into a multi-layered memories that can later be recalled.

Using a special “micro-grid” recording device developed by colleagues at Lawrence Livermore National Laboratory (LLNL), the UCSF researchers were able to measure hippocampus activity in study participants undergoing surgery to treat severe epilepsy. They discovered that brain waves travel back and forth across this structure, integrating messages from different areas of the brain, and showed for the first time what scientists previously had only been able to hypothesize.

A team of researchers from the NSF NHERI SimCenter, a computational modeling and simulation center for the natural hazards engineering community based at the University of California, Berkeley, have developed a suite of tools called BRAILS — Building Recognition using AI at Large-Scale — that can automatically identify characteristics of buildings in a city and even detect the risks that a city’s structures would face in an earthquake, hurricane, or tsunami.

Charles Wang, a postdoctoral researcher at the University of California, Berkeley, and the lead developer of BRAILS, says the project grew out of a need to quickly and reliably characterize the structures in a city.

“We want to simulate the impact of hazards on all of the buildings in a region, but we don’t have a description of the building attributes,” Wang said. “For example, in the San Francisco Bay area, there are millions of buildings. Using AI, we are able to get the needed information. We can train neural network models to infer building information from images and other sources of data.”

Developers are benefitting from big investments over the past several years in research, development, and manufacturing of integrated photonics. That has fertilized the soil for a wave of integrated-photonics startups aiming at the AI market.

In March 2021, a Massachusetts Institute of Technology (MIT; Cambridge, MA) spinoff called Lightmatter (Boston, MA) introduced what the company calls “the world’s first general-purpose photonic artificial intelligence accelerator.” Named Envise, it’s a photonic module that can be plugged into leading models of AIs and neural networks to supercharge their performance, which in the AI world is measured in “inferences” per second. Envise includes an array of 16 Lightmatter photonic chips, and is built into a 4-U server blade format. It has 1 TB of DRAM, 3 TB of solid-state memory, and handles optical interconnections up to 6.4 Tbit/s. As shown in the schematic in Figure 2, the Envise also includes two lasers driving the photonic processors. The basic building blocks are arrays of identical MZIs assembled for matrix multiplication.

The University of Texas at San Antonio now has someone to lead its new School of Data Science that will help anchor a nearly 170,000-square foot structure west of Interstate 35 — the centerpiece of the institution’s planned downtown expansion.

David Mongeau will be the founding director for the school, the first of its kind in Texas.

Mongeau is well recognized for his success leading data science and analytics research institutes and training programs, as well as his ability to develop collaborative partnerships across government, industry, academia and the philanthropic community. He most recently was executive director of the Berkeley Institute for Data Science.

Howard University is collaborating with Amazon Web Services (AWS) to upskill its students and build pathways to technical careers with cutting-edge cloud computing courses and training resources for educators. Through this effort, students will have access to curriculum aligned to in-demand cloud careers, and developed in coordination with AWS experts to prepare students for the workforce, including hands-on experiences to help ensure alignment with the needs of employers seeking skilled cloud talent.

Howard University will integrate cloud computing concepts from AWS Educate into new and existing higher education curriculum, giving students access to foundational cloud computing courses, and advanced technical topics such as machine learning and computer vision.

Apple today announced powerful software features designed for people with mobility, vision, hearing, and cognitive disabilities. These next-generation technologies showcase Apple’s belief that accessibility is a human right and advance the company’s long history of delivering industry-leading features that make Apple products customizable for all users.

Later this year, with software updates across all of Apple’s operating systems, people with limb differences will be able to navigate Apple Watch using AssistiveTouch; iPad will support third-party eye-tracking hardware for easier control; and for blind and low vision communities, Apple’s industry-leading VoiceOver screen reader will get even smarter using on-device intelligence to explore objects within images. In support of neurodiversity, Apple is introducing new background sounds to help minimize distractions, and for those who are deaf or hard of hearing, Made for iPhone (MFi) will soon support new bi-directional hearing aids.

Twitter started using a saliency algorithm in 2018 to crop images. We did this to improve consistency in the size of photos in your timeline and to allow you to see more Tweets at a glance. The saliency algorithm works by estimating what a person might want to see first within a picture so that our system could determine how to crop an image to an easily-viewable size. Saliency models are trained on how the human eye looks at a picture as a method of prioritizing what’s likely to be most important to the most people. The algorithm, trained on human eye-tracking data, predicts a saliency score on all regions in the image and chooses the point with the highest score as the center of the crop.

In our most recent analysis of this model, we considered three places where harms could arise:
1. Unequal treatment based on demographic differences: People on Twitter noted instances where our model chose white individuals over Black individuals in images and male-presenting images over female-presenting images. We tested the model on a larger dataset to determine if this was a problem with the model.
2. Objectification biases, also known as “male gaze”: People on Twitter also identified instances where image cropping chose a woman’s chest or legs as a salient feature. We tested the model on a larger dataset to determine if this was a systematic flaw.

In the midst of a devastating global pandemic of wildlife origin and with future spillovers imminent as humans continue to come into closer contact with wildlife, infectious-disease models that consider the full ecological and anthropological contexts of disease transmission are critical to the health of all life. Existing models are limited in their ability to predict disease emergence, since they rarely consider the dynamics of the hosts and ecosystems from which pandemics emerge.

Published May 17 in Nature Ecology and Evolution, Smithsonian scientists and partners provide a framework for a new approach to modeling infectious diseases. It adapts established methods developed to study the planet’s natural systems, including climate change, ocean circulation and forest growth, and applies them to parasites and pathogens that cause disease.

With Jessica Hullman and Nicholas Diakopoulos, I conducted a qualitative thematic analysis of a sample of 300 NeurIPS 2020 broader impact statements (here’s the dataset). Our analysis surfaces several themes around what authors focus on and how they do so. We organized these themes into a framework with dimensions and sub-dimensions of variation (see the table below).

Broadly, we find that authors write about Impacts and Recommendations for how to mitigate negative consequences and ultimately realize better downstream outcomes.

AI has long been quietly embedding itself into higher education in ways like these, often to save money — a need that’s been heightened by pandemic-related budget squeezes.

Now, simple AI-driven tools like these chatbots, plagiarism-detecting software and apps to check spelling and grammar are being joined by new, more powerful — and controversial — applications that answer academic questions, grade assignments, recommend classes and even teach.

The newest can evaluate and score applicants’ personality traits and perceived motivation, and colleges increasing are using these tools to make admissions and financial aid decisions.

As the presence of this technology on campus grows, so do concerns about it.

Local news organizations still haven’t tapped the technology’s full potential. Lacking the knowledge to approach the use of AI as part of a coherent plan and strategy, adoption of AI technologies at local news organizations remains nascent and experimental. This creates two missed opportunities: the short-term benefit of using existing AI tools to support their organization’s business and reporting, and the long-term benefit that comes from building a staff and infrastructure capable of harnessing these tools.

Today, we’re announcing a new, $3 million initiative to help local news organizations expand the application of artificial intelligence and harness it for their long-term sustainability. Over the next two years, we want to ensure local organizations can develop customized strategic plans to enhance their business sustainability through AI/machine learning; assess their internal AI readiness across various editorial and business lines; and collaborate with trusted AI experts and partners on product innovations.

Asian students and faculty have long been a cornerstone of science in the U.S., drawn by the promise of collaboration and cutting-edge research. However, the Asian community is facing increased racist attacks and scrutiny from the government. A cover story in Chemical & Engineering News, the weekly newsmagazine of the American Chemical Society, explores how Asian scientists are reassessing their futures in the U.S.

In the wake of the COVID-19 pandemic, racist attacks against the Asian community in the U.S. have increased notably, with nearly 4,000 incidents reported between March 2020 and February 2021. Some attribute this to former President Donald Trump’s rhetoric about the pandemic originating in China, writes Senior Editor Andrea Widener. Beyond the pandemic, the Trump administration implemented policies that hinder collaborations with researchers in China, including the China Initiative launched by the U.S. Department of Justice in 2018. Initiatives like this have placed Asian researchers under increased scrutiny and suspicion, and many are still in place despite the new presidential administration. Even before Trump and the coronavirus, Asian scientists have said they struggled with feeling welcome in the U.S. due to stereotypes and perceived cultural differences.

Online May 28, starting at 11 a.m. [registration required]