Data Science newsletter – December 1, 2020

GROUP CURATION: N/A

Powerful new artificial intelligence (AI) technology has demonstrated an ability to match the skills and capabilities of experienced inspectors. Scratches and blemishes that were once difficult for automated systems to capture can now be identified even without the use of samples or adjustment. AI-based image filters are capable of identifying defects on any product background, thereby supporting flexibility on automated production lines.

Characteristics that result in image variation—such as appearance, lighting, surface, deviation from center, and the like—tend to confuse conventional image processing technology. Selection of a feature value and threshold adjustment in these systems requires expert-level image processing knowledge, and even then, the tasks are quite time-consuming because they involve a continuing process of trial and error.

The health of the national energy infrastructure has been on everyone’s mind for decades. Managing this behemoth is getting a boost from initiatives based on AI and designed to cover everything from maintenance planning to quality checks to design. And now, two new investments have joined the fight against aging infrastructure.

National Grid Partners (NGP) recently announced it will invest $6 million in two data analytics startups that use artificial intelligence to protect critical infrastructure and ultimately help reduce costs for customers. This brings NGP’s total investments in emerging technology companies to $175 in the last 24 months.

Wild rice makes a tiny exploding sound when it is struck by a cedar knocking stick. A burst, followed by the sounds of rice falling into a canoe below.

“It’s the sound the universe made when it began,” I was told by my long-time teacher. It’s the sound of life beginning, life continuing. Wild rice, what we know as manoomin, is the food that grows on water, the staple that lies at the heart of my people’s culture. It’s what I’ve given my heart, mind, and body to protect against yet another proposed tar sands pipeline, this time Enbridge’s Line 3, set to cut through Ojibwe territory in northern Minnesota.

It’s what sits in my heart as I push through room after room of decision-makers, legislators, financiers, corporate representatives, fellow advocates, climate scientists, climate deniers, and the rest of the cast of characters in the so-called environmental movement. Here, our sacred manoomin becomes a number, a statistical data point. The land that sustains every life on earth becomes a sum of degrees Celsius, carbon emissions, forest acreage, and economic impacts. Water is reduced from our literal lifeblood to a policy concern, a partisan issue up for debate.

The language of climate is part of the distancing we’ve broadly internalized, as far as I can tell. It’s a piece of the world full of invisible barriers and entrenched pathologies. The story of our self-destruction and what to do has been mostly told in cold, statistical analysis recited by a handful of mostly male, mostly non-POC, almost entirely non-Indigenous voices. The language of land is largely absent or relegated to the category of pitiable platitudes.

The Supreme Court on Monday indicated serious reservations about the ambiguity and scope of the nation’s only major cybercrime law, hinting it may narrow the law’s applicability to avoid criminalizing such acts as checking social media at work.

During arguments in a case involving a Georgia police officer convicted of violating the 1986 Computer Fraud and Abuse Act by accessing a license plate database, the justices pushed a Justice Department lawyer to explain how a ruling in the government’s favor wouldn’t open the door to prosecutions of innocuous behavior. Those could include browsing Instagram on a work computer or performing public-spirited security research to test a system for vulnerabilities.

As our work has progressed it has become clear that we cannot separate the question of algorithmic bias from the question of biased decision-making more broadly. The approach we take to tackling biased algorithms in recruitment, for example, must form part of, and be consistent with, the way we understand and tackle discrimination in recruitment more generally.

A core theme of this report is that we now have the opportunity to adopt a more rigorous and proactive approach to identifying and mitigating bias in key areas of life, such as policing, social services, finance and recruitment. Good use of data can enable organisations to shine a light on existing practices and identify what is driving bias. There is an ethical obligation to act wherever there is a risk that bias is causing harm and instead make fairer, better choices.

Proteins are essential to life, supporting practically all its functions. They are large complex molecules, made up of chains of amino acids, and what a protein does largely depends on its unique 3D structure. Figuring out what shapes proteins fold into is known as the “protein folding problem”, and has stood as a grand challenge in biology for the past 50 years. In a major scientific advance, the latest version of our AI system AlphaFold has been recognised as a solution to this grand challenge by the organisers of the biennial Critical Assessment of protein Structure Prediction (CASP). This breakthrough demonstrates the impact AI can have on scientific discovery and its potential to dramatically accelerate progress in some of the most fundamental fields that explain and shape our world.

We investigate SAT scores by race using the College Board’s
publicly available data for over 2.1 million 2020 high school graduates, with a particular focus on the math section. (This analysis builds on our earlier work on this issue from 2017, “Race gaps in SAT scores highlight inequality and hinder upward mobility.”)

The class of 2020 averaged a score of 523 of 800 on the math section of the SAT, slightly below the College Board’s college-readiness benchmark score of 530. (The College Board predicts that the average SAT test taker will earn less than a C in their first-year math course.) The average scores for Black (454) and Latino or Hispanic students (478) are significantly lower than those of white (547) and Asian students (632). The proportion of students reaching college-readiness benchmarks also differs by race. Over half (59%) of white and four-fifths of Asian test takers met the college readiness math benchmark, compared to less than a quarter of Black students and under a third of Hispanic or Latino students. As we show, there are similar patterns for English, but the gaps are not as stark.

In mid-November, as the United States set records for newly diagnosed COVID-19 cases day after day, the hospital situation in one hard-hit state, Wisconsin, looked concerning but not yet urgent by one crucial measure. The main pandemic data tracking system run by the Department of Health and Human Services (HHS), dubbed HHS Protect, reported that on 16 November, 71% of the state’s hospital beds were filled. Wisconsin officials who rely on the data to support and advise their increasingly strained hospitals might have concluded they had some margin left.

Yet a different federal COVID-19 data system painted a much more dire picture for the same day, reporting 91% of Wisconsin’s hospital beds were filled. That day was no outlier. A Science examination of HHS Protect and confidential federal documents found the HHS data for three important values in Wisconsin hospitals—beds filled, intensive care unit (ICU) beds filled, and inpatients with COVID-19—often diverge dramatically from those collected by the other federal source, from state-supplied data, and from the apparent reality on the ground.

In the swift and chaotic pivot to virtual test-taking, companies like Respondus — along with competitors including Honorlock, ProctorU and Proctorio — have stepped in to help schools keep watch on students. Because the new digital tools are required in certain courses, students are being forced to subject themselves to surveillance inside their own homes and open themselves up to disputes over “suspicious activities,” as defined by an algorithm.

According to the company website, Respondus Monitor uses “powerful analytics … to detect suspicious behaviors during an exam session,” and then flags such behaviors for professors to review once the session concludes.

At the self-described “heart” of the company’s monitoring software is Monitor AI, a “powerful artificial intelligence engine” that collects facial detection data and keyboard and mouse activity to identify “patterns and anomalies associated with cheating.”

Privacy advocates have been raising alarms about this type of technology and how easily it’s infiltrating the lives of students during a public health emergency.

The two sides of the COVID-19 war are illustrated by two documents, the John Snow Memorandum and the Great Barrington Declaration, which were posted online in October. The former represents the majoritarian position, which supports strict measures to limit human contacts and movements across the board. In the U.S., this approach has included shutting down public gatherings, restricting hotels and numerous small businesses such as restaurants and shops, and throwing millions of Americans out of work. The Great Barrington Declaration, the minority view, advocates “focused protection,” allowing younger and healthier individuals to continue life, work and going to school, while aiming more protective measures at the most vulnerable to the virus—the elderly, the institutionalized and other high-risk individuals. This approach has been employed most fully in Sweden.

Advocates on both sides have dug in, hurling dismissive and vitriolic attacks at individuals in the other camp. These attacks have caused serious reputational harm in some cases and led at least a few scientists to self-censor and avoid publishing data that could inform efforts to dampen death rates from both the virus and its remedies.

A startup artificial intelligence company with roots at the University of Massachusetts Amherst recently received a coveted two-year, $750,000 Phase II grant from the National Science Foundation’s (NSF) Small Business Innovation Research program – also known as “America’s Seed Fund” – to advance their research on curbing disinformation online.

AuCoDe’s CEO Shiri Dori-Hacohen says, “Conspiracy theories, bots and state-sponsored disinformation campaigns show how crucial factual information is in today’s chaotic climate that includes the pandemic, civil unrest and the intensely divisive U.S. election cycle. NSF’s continued support for AuCoDe’s research recognizes that our patent-pending machine learning approach and strongly qualified team is uniquely positioned to improve the health and veracity of the entire online ecosystem.”

Computer scientist Shiri Dori-Hacohen founded AuCoDe – for automated controversy detection – in 2016 as a spinoff of UMass Amherst computer science NSF-funded research on online controversies with co-founder and computer science professor James Allan, a world-renowned expert in information retrieval. Serial entrepreneur Julian Lustig-Gonzalez joined the team as business co-founder in 2019.

The Center for Exascale Simulation of Materials in Extreme Environments (CESMIX) — based at the Center for Computational Science and Engineering (CCSE) within the MIT Stephen A. Schwarzman College of Computing — will bring together researchers in numerical algorithms and scientific computing, quantum chemistry, materials science, and computer science to connect quantum and molecular simulations of materials with advanced programming languages, compiler technologies, and software performance engineering tools, underpinned by rigorous approaches to statistical inference and uncertainty quantification. … MIT was among a total of nine universities selected as part of the Predictive Science Academic Alliance Program (PSAAP) III to form a new center to support science-based modeling and simulation and exascale computing technologies. This is the third time that PSAAP centers have been awarded by the U.S. Department of Energy’s National Nuclear Security Administration (DoE/NNSA) since the program launched in 2008 and is the first time that the Institute has ever been selected. MIT is one of just two institutions nationwide chosen to establish a Single-Discipline Center in this round and will receive up to $9.5 million in funding through a cooperative agreement over five years.

The Tufts School of Engineering will launch an online Master of Science in Data Science program and an online Postbaccalaureate in Computer Science program in September 2021 to meet the growing demand for coding skills and offer a flexible way for working professionals and students to advance their careers in data science and computer science.

The School of Engineering collaborated with Noodle Partners, an online program management company, to develop the new programs and course material.

During the early days of the pandemic, whether or not you had a backbone could have been the difference between life and death. At least if you were a lab animal.

“(Oregon State University) would let you take care of fish during the shutdown, but invertebrates were not considered to be, I guess, important enough,” said OSU eco-toxicologist Susanne Brander.

She studies how microplastics and chemical pollutants affect sea life – including the lowly spineless mysid shrimp that plays a big role in the ocean food web.

Not being able to access her lab was a big problem, so she had to come up with a solution on the fly.

“The best place to put them was my basement,” she said.

“Any intervention that minimizes this travel back and forth… it’s the right thing,” Dr. Peter Chin-Hong, who specializes in infectious diseases at the University of California San Francisco, told Yahoo Finance. “The whole country is on fire, basically, with COVID right now.”

After spring break, according to a new study that analyzed smartphone location data, the return of students to various college towns was followed by “an increase in the growth rate of confirmed cases during the two weeks following the end of spring break.”

“Estimates imply that counties with more early spring break students had a higher growth rate of cases than counties with fewer early spring break students,” authors Daniel Mangrum of the Federal Reserve Bank of New York and Paul Niekamp of Ball State University wrote. “The increase in case growth rates peaked two weeks after spring break.”

Online December 7, starting at 5 p.m. “COVID-19 has caused a staggering 1.26 million deaths worldwide. Nearly a year to the date of when the virus was first reported internationally, science journalist Ed Yong speaks about the global pandemic as part of the 2020 John Jay Iselin Memorial Lecture.” [registration required]

Online December 8, starting at 4 p.m., organized by Radcliffe Institute for Advanced Study at Harvard University. “To what extent is our future with COVID-19 knowable? As new information about the transmission, demographics, and treatment of COVID-19 emerge, epidemiologists continue to address complex data and generate new predictive models to better understand the dynamics of the virus. Join leading epidemiologists for a panel discussion as they assess the current and future state of the epidemic.” [registration required]

“Faceted search has become a critical feature for enhancing findability and the user search experience for all types of search applications. This article gives you an introduction to faceted search with Apache Solr.”

This is the second of our three-part series, Exploring The Intersection Of Design & Advanced Analytics. In this article, we will take a deeper look at how effective collaboration between designers and data scientists can unlock greater value in advanced analytics (AA) projects.