Data Science newsletter – June 5, 2021

Fueled by generous philanthropic support and insights from industry partners, the University of Delaware launched a new master’s program in biopharmaceutical sciences in June. … “It’s not often that we are part of establishing a program like this,” said Michael Thien, senior vice president at Merck and member of UD’s Biopharmaceutical Innovation Advisory Board. “This has given us the opportunity to provide a lot of input as to the relevant experiences and knowledge we are looking for in the students as we bring them on board after graduation. We can ensure that we are teaching these students not only about where the biopharma industry is now, but where it’s going. One of the areas that will be really helpful for the students to learn about is data science: how to address ‘big data’ questions and analyze very large streams of data. Since the biotech revolution in the late 80s, changes were largely incremental. But now we’re seeing the industry take a huge step forward and students need to be ready to take on this new world that includes machine learning and artificial intelligence for data processing. They are digital natives that grew up in a digital world, but they will need to make sure they understand the fundamentals of data analytics and know what to do with it.”

We have gobs of data, nearly limitless cloud compute, and ever-improving machine learning algorithms, so what on earth is holding companies back from succeeding with big data? “Talent, talent, talent,” says Dr. Kirk Borne. “The limiting factor is talent.” … “There’s almost exponential growth in the talent pool,” Borne tells Datanami. “But unfortunately, so to speak, for the business world, the number of job opportunities that businesses are creating is also exponentially growing, but at a faster pace than the talent production. The difference between two exponentials is effectively still an exponential, and so there’s still this rapidly growing talent gap.”

The US National Science Foundation appears set to receive a large funding increase and a new technology directorate in the near future—although just how much money will be involved remains to be worked out among the House of Representatives, Senate, and White House.

All those bodies have proposals in the works to beef up the NSF, which funds about 25% of all federally funded basic research in the US. But its budget—$8.5 billion for fiscal 2021—has had only minimal increases over the past several years—far smaller on a percentage basis than the National Institutes of Health, for example.

“NSF is a sort of beacon for all innovation in the country,” says biological engineer Subra Suresh, who was the NSF director from 2010 to 2013 and is now president of Nanyang Technological University. “From that perspective, an increase to NSF funding is long overdue.”

Purdue University is offering a new all-online master’s degree in data science in finance with a concentrated curriculum focus on machine learning to solve modern financial problems.

The online degree offers flexibility for working professionals in the financial industry who want to advance their skills and careers, and it makes the Purdue master’s in data science in finance readily available to an international audience of students.

It seemed like Big Tech, with its analytic firepower and new focus on health, could help with these very real problems. “We saw all over the papers: Facebook is gonna save the world, and Google’s going to save the world,” says Katerini Storeng, a medical anthropologist who studies public-private partnerships in global public health at the University of Oslo. Politicians were eager to welcome Silicon Valley to the table and to discuss the best ways to manage the pandemic. “It was remarkable, and indicative of a blurring of the boundaries between the public domain and the private domain,” Storeng says.

Over a year later, many of the promised tech innovations never materialized. There are areas where tech companies have made significant contributions — like collecting mobility data that helped officials understand the effects of social distancing policies. But Google wasn’t actually building a nationwide testing website. The program that eventually appeared, a testing program for California run by Google’s sibling company Verily, was quietly phased out after it created more problems than it solved.

Now, after a year, we’re starting to get a clear picture of what worked, what didn’t, and what the relationship between Big Tech and public health might look like in the future.

The Supreme Court issued a decision today in Van Buren v. United States, narrowly interpreting the Computer Fraud and Abuse Act (CFAA) in a manner protecting civil rights researchers and journalists who test for discrimination online. The court’s decision makes clear that violations of website terms of service alone are not enough to create liability under the CFAA.

The ACLU has long challenged the interpretation of the CFAA that would create criminal and civil liability for researchers and journalists who violate website terms of service while undertaking civil rights testing and research online. The ACLU filed its challenge to the constitutionality of the CFAA, Sandvig v. Barr, in 2016, on behalf of academics and journalists seeking to test websites for discrimination among users. In 2020, the district court held that the CFAA should not be read to prohibit violations of website terms of service alone.

In today’s decision, the Supreme Court cited the ACLU’s amicus brief on behalf of civil rights testers, as well as government testimony elicited in the Sandvig district court win.

There is a growing concern that e-commerce platforms are amplifying vaccine-misinformation. To investigate, we conduct two-sets of algorithmic audits for vaccine misinformation on the search and recommendation algorithms of Amazon—world’s leading e-retailer. First, we systematically audit search-results belonging to vaccine-related search-queries without logging into the platform—unpersonalized audits. We find 10.47% of search-results promote misinformative health products. We also observe ranking-bias, with Amazon ranking misinformative search-results higher than debunking search-results. Next, we analyze the effects of personalization due to account-history, where history is built progressively by performing various real-world user-actions, such as clicking a product. We find evidence of filter-bubble effect in Amazon’s recommendations; accounts performing actions on misinformative products are presented with more misinformation compared to accounts performing actions on neutral and debunking products. Interestingly, once user clicks on a misinformative product, homepage recommendations become more contaminated compared to when user shows an intention to buy that product. [full text]

Amazon is building a wireless network – using your internet bandwidth.

It’s called Amazon Sidewalk, and the company touts it as a way to help its devices work better, by extending the range of low-bandwidth devices to help them stay online.

It does that by pooling neighbors’ bandwidth to help connectivity for devices that are out of range.

The network already includes certain Ring Floodlight Cam and Spotlight models as of late last year. Compatible Amazon Echo devices will be added to the network on Tuesday. Ring Doorbell Pro devices will be able to access Sidewalk, too.

Amazon intends to defend itself against claims that it violated an Illinois biometric privacy statute by arguing that the law is unconstitutional.

That’s according to documents Amazon filed with U.S. District Court Judge James Robart in Seattle late last month.

The tech company outlined numerous planned defenses to the claims, including that an unfavorable verdict would effectively “abridge Amazon’s freedom of speech, in violation of the First Amendment.”

On May 21, 2021, the U.S. Consumer Products Safety Commission (“CPSC”) published a report on artificial intelligence (AI) and machine learning (ML) in consumer products. The report highlights recent CPSC staff activity concerning AI and ML, proposes a framework for evaluating the potential safety impact of AI and ML capabilities in consumer products, and makes several recommendations that the CPSC can take in identifying and addressing potential hazards related to AI and ML capabilities in consumer products.

Concerning staff activity, CPSC recently hired a Chief Technologist with a background in AI and ML to address the use of AI in consumer products. The CPSC also recently established an “AI/ML Working Group” and held a virtual forum on AI and ML in March 2021.

Informed by the discussions held with various stakeholders at this forum, the CPSC staff has proposed a framework in the report for evaluating the potential safety impact of AI and ML in consumer products.

But employees are finding out the paper was funded by FB.
https://papers.ssrn.com/sol3/papers.cf

The discipline of computer science has historically made effective use of peer-reviewed conference publications as an important mechanism for disseminating timely and impactful research results. Recent attempts to “game” the reviewing system could undermine this mechanism, damaging our ability to share research effectively.

I want to alert the community to a growing problem that attacks the fundamental assumptions that the review process has depended upon. My hope is that exposing the behavior of a community of unethical individuals will encourage others to exert social pressure that will help bring colluders into line, invite a broader set of people to engage in problem solving, and provide some encouragement for people trapped into collusion by more senior researchers to extricate themselves and make common cause with the rest of the community. My motivation for writing this Viewpoint is because I became aware of an example in the computer-architecture community where a junior researcher may have taken his own life instead of continuing to engage in a possible collusion ring.a

Collusion rings extend far beyond the field of computer architecture. I will share another data point, from artificial intelligence and machine learning. I will keep some of the details (like the identity of the specific conference) vague because I think naming names could do more harm than good. Since my goal is to raise awareness of the issue and help people understand how widespread it is, I do not think such details are essential.

There will surely never be a return to the old days, when platforms such as Facebook and Twitter tried to wash their hands of the bulk of what happened on their sites with faith that internet users, as a global community, would magically govern themselves. But a slow and steady march toward a future where ever more problems are sought to be addressed by trying to erase content from the face of the internet is also a simplistic and ineffective approach to complicated issues. The internet is sitting at a crossroads, and it’s worth being thoughtful about the path we choose for it. More content moderation isn’t always better moderation, and there are trade-offs at every step. Maybe those trade-offs are worth it, but ignoring them doesn’t mean they don’t exist.

A look at how we got here shows why the solutions to social media’s problems aren’t as obvious as they might seem. Misinformation about the pandemic was supposed to be the easy case. In response to the global emergency, the platforms were finally moving fast and cracking down on Covid-19 misinformation in a way that they never had before. As a result, there was about a week in March 2020 when social media platforms, battered by almost unrelenting criticism for the last four years, were good again. “Who knew the techlash was susceptible to a virus?” Steven Levy asked.

Healthcare AI has potential not only for neutralizing its inherent algorithmic bias but also for personalizing its outputs to help humans address health inequities.

All it will take is incorporating an ingredient missing from most algorithm training: patient input.

In fact, early testing of this hypothesis has already demonstrated the feasibility of patient-inclusive AI development.

Specifically, some of this research “has produced an algorithm trained to predict the knee pain reported by the patient, rather than the X-ray interpretation of the doctor,” the authors of a May 29 Lancet opinion piece write.

Walgreens is leveraging customer data to boost equitable access to medicines for diabetes, asthma, and other illnesses as part of a broader industry effort to close the gap in health care for underserved communities.

The company has started using a data tracking model to identify and reach out to people who are most at risk, such as in communities of color or low-income areas. The model enables pharmacists to take a holistic approach to meeting the immediate needs of their customers and identify best practices, according to Michael Taitel, a senior director of health analytics, research & reporting at Walgreens.

“Our purpose is to champion the health and well-being of every community in America, and that doesn’t mean where it’s easy,” he said Wednesday during a webinar on health equity. “It means we need to put in extra effort and partner with communities that are experiencing health disparities.”

Online June 15, starting at 7 p.m. Central. “It’s so good to reconnect with the @eyeofestival
community – even on screens. Jer & I will catch up on AI, politics, art, and The Discourse™. Join us!”

London, England September 14-16. “Data for Policy is a premier global forum for interdisciplinary and cross-sector discussions around the impact and potentials of the digital revolution in the government sector.”

Requirements Engineering (RE) is a critical area in software development, as figuring out what to develop and include in a product is a cornerstone activity which all others depend upon. Countless studies of unsuccessful development projects report that lack in RE is often a core-contributing failure factor.13 Central in RE is the role that coordinates all its related activities, usually named requirements engineer. Still, empirical evidence on the way companies implement this role is scarce. In this article, we present the results of an interview-based descriptive study involving 24 IT professionals from 12 companies. As a main outcome, we can affirm that all companies assign IT professionals to the requirements engineer role in their projects, but in many different ways, which might impact efficiency of the function. Furthermore, we uncover that requirements engineers often perform other tasks ranging from project’s go vs. no-go decisions to test suite design in addition to handling requirements. Last, the study highlights their need to communicate with many other roles inside the company, from domain experts to system architects.

INSTRUCTORS Yann LeCun & Alfredo Canziani

Check the repo’s README.md and learn about:

Content new organisation

The semester’s second half intellectual dilemma