Data Science newsletter – October 14, 2019

GROUP CURATION: N/A

Ramapo College of New Jersey has launched two new degree programs in data science.

The Mahwah-based higher educational institution said on Wednesday it has created a Bachelor of Science and a Master of Science for the program, effective Fall 2020.

The University of Kansas has been awarded a $3 million grant from the National Science Foundation to develop an innovative graduate training program that combines the disciplines of chemistry, chemical engineering and computer science.

“It’s a big win for the university,” said Kevin Leonard, associate professor in the School of Engineering’s Department of Chemical & Petroleum Engineering. He will lead the new program for KU’s Center for Environmentally Beneficial Catalysis.

In 2019, the war for ML frameworks has two remaining main contenders: PyTorch and TensorFlow. My analysis suggests that researchers are abandoning TensorFlow and flocking to PyTorch in droves. Meanwhile in industry, Tensorflow is currently the platform of choice, but that may not be true for long.

In an interview with Neuron, Dr. John Krakauer talks about his lab’s current work on motor learning and neurorehabilitation and shares his views on promoting “the slow and deep over the fast and shallow” and on the crucial need for diversity in academia.

Surging student demand for data science skills and an abundant supply of jobs for people with related expertise have led UW-Madison to create a new school, add an academic program and hire more faculty focused on data science.

After years of strained resources for individual academic departments teaching technology skills, the university’s new School of Computer, Data & Information Sciences is a concentrated, coordinated effort to equip more students with what college officials see as required tools to work in the modern world.

“A hundred years ago, you may have thought of mathematics as that kind of basic tool that everybody needs to go study other fields,” said Department of Computer Sciences chairman Remzi Arpaci-Dusseau. “Now we believe that computation, data, and then being ethical in how you apply those things, are that next level of foundation.”

Data is proliferating at warp speed, but data literacy among managers and executives hasn’t caught up. That’s why data translators have, according to the Harvard Business Review, become a “must-have analytics role.” Even if you’ve never heard the term data translator, you may already be working with one. Because they go by so many different titles—like MassMutual’s product managers—no one knows how many translators exist right now. But there’s no doubt that people who are adept at interpreting data for practical use in the real world are a hot commodity. By 2026, the McKinsey Global Institute predicts that there could be a demand for 2 million to 4 million translators in the U.S. alone.

At the moment, hiring translators isn’t easy. That’s partly because the job requires a unique combination of skills, usually including both a strong grounding in data science and a talent for boiling complex ideas down to clear, practical choices. They’re so rare that translators “belong to a category recruiters call ‘unicorns’,” notes Brad Stillwell, vice president of product strategy at Birst, a unit of global cloud software giant Infor.

London design studio Sample & Hold has been asked to scan all kinds of things: a shoe, a carrot, the heads of every member of the Barcelona FC team. … Sample & Hold doesn’t need lasers to do this 3D scanning. Instead, it uses plain old 2D cameras. The trick is to use lots of them – 67 in total.

Poaching is one of the biggest threats to the survival of the species. The parts from a single tiger can fetch more than $50,000 on the black market, and organized gangs of poachers kill dozens of the animals each year. At wildlife preserves throughout Africa and Asia, overstretched and under-resourced park rangers struggle to keep pace with this crisis.

A new tool being developed at the Harvard John A. Paulson School of Engineering and Applied Sciences seeks to give rangers an edge in their battle against poachers. Protection Assistant for Wildlife Security (PAWS) is an artificial intelligence system that predicts poaching risk levels in different areas of a wildlife preserve and helps rangers patrol more efficiently.

UC San Francisco is launching a new center to accelerate the application of artificial intelligence (AI) technology to radiology, leveraging advanced computational techniques and industry collaborations to improve patient diagnoses and care.

The Center for Intelligent Imaging, or ci2, will develop and apply AI to devise powerful new ways to look inside the body and to evaluate health and disease. Investigators in ci2 will team with Santa Clara, Calif.-based NVIDIA Corp., an industry leader in AI computing, to build infrastructure and tools focused on enabling the translation of AI into clinical practice.

We’re living through an extraordinary moment in technological history. In the past decade, the rise of artificial intelligence (both in theory and in practice) has revolutionized computer science and the workplace. As the field expands, many philosophers and academics are raising questions about what A.I. means for the future of human intelligence. This week, we’ve gathered a selection of pieces about the evolution of artificial intelligence and its impact on our lives. In “How Frightened Should We Be of A.I.?,” Tad Friend explores the risks of building systems that are smarter than we are. In “The Doomsday Invention,” Raffi Khatchadourian examines the work of the philosopher Nick Bostrom, who argues that true A.I. may pose a future threat to humanity. Siddhartha Mukherjee writes about the advantages and disadvantages of the use of automation in medical diagnoses, and Sheelah Kolhatkar chronicles the ways in which industrial robots have replaced workers in factories across the country. In “Get Smart,” Adam Gopnik considers how we will know when machine intelligence has eclipsed that of humans. Finally, forty years after the début of the computer HAL, in Stanley Kubrick’s “2001: A Space Odyssey,” John Seabrook looks at humans’ desire, and ability, to talk to computers. We hope that you enjoy these views of an expanding and ever-changing technological frontier.

he University of Alaska Board of Regents will no longer consider merging UA’s three separately accredited public universities into one accredited institution — at least for now.

Regents voted 9-2 at an emergency meeting Monday in Fairbanks to “cease consideration of a single accreditation” until the University of Alaska Fairbanks’ accreditation is reaffirmed in 2021, according to the approved motion, read by Regent Karen Perdue. Regents Dale Anderson and Mary Hughes cast the dissenting votes.

The regents’ decision followed a discussion with the accreditor of Alaska’s universities Monday morning and a nearly two-hour closed-door executive session. It marks a sharp shift for the regents during an especially turbulent year for the public university system.

For Stormy Chamberlain, it wasn’t about the money as much as the title: an endowed assistant professorship.

Endowed positions typically go to faculty members much further along in their careers. Getting one so soon after starting at the University of Connecticut, when she was still an assistant professor of genetics and genome sciences, meant that Chamberlain had something impressive to put on her CV. It helped her land federal funding for her research, she says, which is critical for young academic scientists to advance their careers. She’s now an established researcher, having led three National Institutes of Health-funded studies of rare genetic conditions that affect children at birth.

But the title of Chamberlain’s professorship has taken on a different tenor since she got it, around 2010. It’s the Raymond and Beverly Sackler Endowed Assistant Professorship.

A project to produce detailed maps of all the land on Earth through laser scanning has been revealed by researchers who say action is needed now to preserve a record of the world’s cultural, environmental and geological treasures.

Prof Chris Fisher, an archaeologist from Colorado State University, said he founded the Earth Archive as a response to the climate crisis.

“We are going to lose a significant amount of both cultural patrimony – so archaeological sites and landscapes – but also ecological patrimony – plants and animals, entire landscapes, geology, hydrology,” Fisher told the Guardian. “We really have a limit time to record those things before the Earth fundamentally changes.”

Global modeling of the atmosphere’s chemistry requires immense computational resources to study how particles form and are processed. A team of researchers, led by Randall Martin at Washington University in St. Louis, plans to improve a high-performance model so that it is more accessible and easier to use by researchers worldwide.

Martin, a professor of energy, environmental and chemical engineering in the McKelvey School of Engineering, will work to improve the GEOS-Chem model with a two-year, $1.2 million grant from NASA. He will collaborate with researchers from Harvard University, Massachusetts Institute of Technology and NASA Goddard Space Flight Center to improve the functionality of the high-performance model, known as GCHP.

Deep within a rainforest in Belize, scientists using lasers beamed from an airplane to peer beneath the dense foliage have discovered evidence of a vast network of ancient Maya farms that date back thousands of years.

The findings, part of more than 20 years of research in this part of Central America, show how the ancient Maya civilization, which reached its peak at around 250 A.D. to 900 A.D., adapted their farming practices in the face of environmental challenges.

San Francisco, CA October 29-November 1. “ODSC is comprehensive and totally community-focused: it’s the conference to engage, to build, to develop, and to learn from the whole data science community.” [$$$]

I find both the Library Of Babel and GPT-2 fascinating. They are both related to language and computer generated text. But hitting “Random” on the Library Of Babel will result in a page filled with gibberish with a probability indistinguishable from 100%. This gave me the idea that one could try to improve this by using a machine language model such as GPT-2.

If GPT-2 “knows” what is the most likely word to follow then one could use this ability to make “real” texts more accessible to a universal librarian. For that, the first thing to do is to replace letter generation by word generation. This is because GPT-2 generates words and not letters. Also thanks to that, a random query will return only actual words rather than unpronounceable nonsense — but there is a trade off. In doing so, we prevent the system to produce texts with non-existing words ((non)-existing here is taken is the liberal sense that a word would have or not been learned by the model in use).

The second idea is to make shorter indexes for pages with the “most likely” content, that is, the content that GPT-2 predicts is the most likely. For that one needs to modify the way the system generates its predicted text so that it is possible to select a word according to its rank of likelihood.

Whether you’re a machine learning enthusiast, researcher, or professional, setting up and managing your work environment can be a complex and distracting process. Even when you’re done dealing with installation and version compatibility issues, you’ll face challenges around cloning the latest models, sharing your results with others, and simply keeping track of your work. To make matters worse, many cloud GPUs are prohibitively expensive.

We are addressing these issues head-on with the announcement of Gradient Community Notebooks, a free cloud GPU service based on Jupyter notebooks that radically simplifies the process of ML/AI development.

The problem – apart from Perl’s dwindling popularity – simply is that Perl 5 and Perl 6 are separate, but related, programming languages. And when developers talk about Perl, they usually mean Perl 5. To give Perl 6 a chance to thrive on its own, the Perl community mostly agreed that a new name would be helpful. … Ultimately, Raku, a reference to Perl 6’s Rakudo compiler, won out. Perl 6 will become Raku, assuming the four people who haven’t yet approved the pull request give their okay, and Perl 5 will become simply Perl.