Data Science newsletter – January 19, 2021

GROUP CURATION: N/A

First, what doesn’t work: shallow, showy diversity efforts (even if they are well-intentioned) aren’t just ineffective, they actively cause harm.

Spend time thinking through your strategy & making sure you can back it up 2/

For example, if you start a “women & allies” email list and then fire a Black woman for being honest on it, it probably would have been better not to have the email list in the first place 3/

In the past ten years, new data science courses and programs have proliferated at the collegiate level. As faculty and administrators enter the race to provide data science training and attract new students, the road map for teaching data science remains elusive. In 2019, 69 college and university faculty teaching data science courses and developing data science curricula were surveyed to learn about their curriculum, computing tools, and challenges they face in their classrooms. Faculty reported teaching a variety of computing skills in introductory data science (albeit fewer computing topics than statistics topics), and that one of the biggest challenges they face is teaching computing to a diverse audience with varying preparation. The ever-evolving nature of data science is a major hurdle for faculty teaching data science courses, and a call for more data science teaching resources was echoed in many responses. [full text]

I’ve spent two decades as a researcher and educator in artificial intelligence, drawn to the field by the opportunity to explore the mysteries of perception and cognition. But life is rarely as simple as we’d like, and the arc of my career has paralleled my mother’s escalating health struggles, including a chronic, life-threatening cardiovascular condition. As all-consuming as the world of academia can be, it sometimes feels as if I’ve spent as much time in hospitals as I have in my lab.
advertisement

I’m happy to report my mother continues to persevere, but her resilience hasn’t been the only silver lining to this ordeal. Years spent in the company of nurses and doctors—unfailingly committed, but perpetually overworked and often sleep deprived—convinced me that the power of AI could radically elevate the way care is delivered. Intelligent sensors could keep tireless watch over patients, automate time-consuming tasks like charting and transcription, and identify lapses in safety protocols as they happen. After all, if AI can safely guide cars along freeways at 70 miles per hour, I wondered, why can’t it help caregivers keep up with the chaos of the healthcare environment?

At the heart of this idea was an obstacle, however. I was proposing research that extended beyond the limits of computer science and into an entirely different field, with decades of literature and traditions stretching back generations. It was clear I needed a collaborator—not just an authority in healthcare, but one with the patience and open-mindedness to help an outsider bring something new to the table. For the first time in my career, success would depend on more than the merits of my work; it would require the humility of researchers like me to recognize the boundaries of our knowledge, and the graciousness of experts in another discipline to help us overcome them.

“Don’t Build a High-End Gaming PC Right Now,” Tom’s Hardware declared in November. “This Is a Bad Time to Build a High-End Gaming PC,” ExtremeTech echoed last month. But it’s looking like the worst time to build a new gaming rig is still to come. That’s because a perfect storm is pushing the price of components like must-have Nvidia or AMD graphics cards even further skyward — and because Donald Trump is now part of that storm.

The Trump administration is now imposing a 25 percent tax on graphics cards imported from China, and at least one PC component manufacturer has announced it’ll pass along part of that cost to consumers.

This updated work follows a full study released in November 2020, on faculty hiring in Computer Science for hires starting in 2021. That work analyzed hiring based on ads through mid-November 2020 and found significant decreases in the number of institutions searching and the number of positions being sought. This updated work considers ads through the end of December 2020 and is intended to understand the impact of the COVID-19 pandemic on whether searches have been delayed or simply will not materialize this hiring season.

In this updated work, we analyzed ads from 319 institutions seeking to fill tenure-track faculty positions in Computer Science, which is an increase from 235 institutions as of mid-November 2020. This updated number is still a 29% decrease from last year at this time and the lowest number in five years.

As we stare into the dawn of 2021, children in the United States face a stark reality. Many have been out of classrooms since March 2020. Online educational alternatives are spotty. Grades and achievement scores are slumping, with the biggest losses for the most vulnerable students. An analysis by consultancy McKinsey & Company, based in New York City, estimates that white students could lose 4–8 months of learning in mathematics, and students of colour 6–12. It’s unclear when many US schools should reopen for in-person learning. Tragically, the country still lacks data that show what’s safe. The fragmented data collection in schools echoes the country’s slipshod approach to tracking COVID-19 in care homes and hospitals, and even basic case- and death-rate information.

For any scientific report, repeating the original analyses upon the original data should yield the original outcomes. We evaluated analytic reproducibility in 25 Psychological Science articles awarded open data badges between 2014 and 2015. Initially, 16 (64%, 95% confidence interval [43,81]) articles contained at least one ‘major numerical discrepancy’ (>10% difference) prompting us to request input from original authors. Ultimately, target values were reproducible without author involvement for 9 (36% [20,59]) articles; reproducible with author involvement for 6 (24% [8,47]) articles; not fully reproducible with no substantive author response for 3 (12% [0,35]) articles; and not fully reproducible despite author involvement for 7 (28% [12,51]) articles. Overall, 37 major numerical discrepancies remained out of 789 checked values (5% [3,6]), but original conclusions did not appear affected. Non-reproducibility was primarily caused by unclear reporting of analytic procedures. These results highlight that open data alone is not sufficient to ensure analytic reproducibility. [full text]

From what I could see, Teams hoovers up all your chats, voicemails, shared meetings, files, transcriptions, your profile details including your email address and phone number, and a detailed analysis of what you were wearing on the call. (I may have made up that last one.)

Cut to September and Microsoft offered a little more about the Teams Activity Report (since updated). Here’s a sentence that’s unsurprising but still a touch uncomfortable: “The table gives you a breakdown of usage by user.”

Everything from how many meetings that user organized to how many urgent messages they sent is recorded. Separate numbers are given for scheduled meetings and those that were ad hoc. Even individuals’ screen-share time is there.

The Health and Human Services Department recently selected Oki Mek as its first-ever chief artificial intelligence officer.

“AI is playing and will continue to play a significant role in overall technology modernization,” HHS Chief Information Officer Perryn Ashmore told Nextgov via email Thursday. “As such, I have named Oki Mek the Chief Artificial Intelligence Officer (CAIO) for the Office of the Chief Information Officer.”

Mek enters the role with almost two decades of federal, technology-centered work, according to his LinkedIn profile. Over the last couple years, he served as chief technology officer in the massive health agency’s division of acquisition, and as a senior advisor to the CIO.

Meet the Cakie (half cake/half cookie) and the Breakie (half bread/half cookie).

The debate on digital and robotic “love” is highly polarised, much like most major debates in the history of technology. Usually, consensus is reached somewhere in the middle.

But in this debate, it seems the technology is advancing faster than we are approaching a consensus.

Generally, the most constructive relationship a person can have with technology is one in which the person is in control, and the technology helps enhance their experiences. For technology to be in control is dehumanising.

A trio of new studies demonstrate low risk of COVID-19 infection and spread in schools, including limited in-school COVID-19 transmission in North Carolina, few cases of the coronavirus-associated multisystem inflammatory syndrome in children (MIS-C) in Swedish schools, and minimal spread of the virus from primary school students in Norway.

A bill introduced in the California State Assembly on Thursday excoriates the fast food industry for its response to the pandemic, contending that chains have “routinely flouted” measures intended to protect the state’s 500,000 fast food workers – and their millions of customers – from COVID-19.

It is well known that a neuron is the main cell of the nervous system. It generates electrical pulses as large as 0.1Volts, has sprouts — axons (as long as 1 meter) and dendrites (several millimeters), and lives more than 100 years. You will not find another more marvelous cell in the body. Still, I will not write about biological details (despite how interesting they are). Rather, I will try to define the most important computational principles, which might help with the progression to artificial intelligence (AI).

Convergence towards principles always leads to simplification and generalization, but it can be dangerous. The brain is too diverse. You can simplify too much and lose something important. Let’s dare to try. Ultimately, it is an iterative process. We merely want to become less wrong with time.

A group of Cal Poly faculty from all six colleges is working to support collaborative data-driven, cross-disciplinary research and to expand on-campus educational opportunities in the areas of data science, data analytics and data literacy.

The team hopes to engage a broad group of stakeholders to envision a Cal Poly that embraces “data science for all” in a way that supports innovative faculty research and enhances student development and learning. The team also emphasizes “data for good” as a guiding principle, ensuring that ethical concerns — about the creation and use of data, as well as the imperatives to increase diversity, equity and inclusion — are infused throughout their efforts.

Online February 2, starting at 7 p.m. Eastern time. “A Conversation with Charlton McIlwain, Professor of Media, Culture, and Communication, NYU.”

Data Scientists use to work with notebooks like Jupyter and RMarkdown. Through notebooks, they can easily share their analysis in HTML format. But what about when there is a need to share the notebooks publicly? In this case, the most convenient way is to configure an Amazon S3 bucket to function as a static website. In this tutorial, we will provide you a walkthrough example of how you can share your notebooks as a static website with AWS S3.

How to pipe your nest camera’s feed into a custom trained neural network to detect anything you want. Cars in your garage, backyard raccoon invaders, pooping pooches, sky’s the limit. [video, 23:51]

We also have video of some of the talks from our Data Science Leadership Summit and Annual Meeting last October. Check them out!

A great place to submit papers for those working on what makes us collectively smarter, or dumber. Open for submissions starting today.