NYU Data Science newsletter – April 23, 2015

GROUP CURATION: N/A

… Like personal computing in the 1980’s or the Internet in the 1990’s, today’s data revolution presents a new opportunity to radically transform virtually every field for the better. New data sources and statistical tools can help identify trends, recognize inefficiencies and discover information that leads to greater impact in every area, from healthcare to urban planning. All organizations are now data organizations.

Humanitarian organizations are no exception. They’re inundated with data from the public sector, from the digital tools they use every day and by newer programs that specifically utilize mobile programs or the Internet. Yet many lack the resources to capitalize on these opportunities.

Inside a lab at Facebook, Yael Maguire is building infrared lasers, Earth-orbiting satellites, and a fleet of flying drones powered by the light of the sun. It sounds more like the plot of the latest James Bond movie than the work of a social networking company. But Maguire’s project is a direct path to the future of Facebook—and the Internet as a whole.

Maguire oversees what Facebook calls its Connectivity Lab. With those lasers, satellites, and drones, he and his team of engineers are working to bring the Internet to all those people on Earth who don’t already have it.

Three summers ago, at the age of 64, Geoffrey Hinton left his home in Toronto’s Annex neighbourhood to become an intern at Google. He received a propeller beanie stitched with the word “Noogler” and attended orientation sessions populated mostly by millennials, who seemed to regard him, in his words, as a “geriatric imbecile.”

In fact, Google wanted him because he is the godfather of a type of artificial intelligence currently shattering every ceiling in machine learning.

In January, the British-American computer scientist Stuart Russell drafted and became the first signatory of an open letter calling for researchers to look beyond the goal of merely making artificial intelligence more powerful. “We recommend expanded research aimed at ensuring that increasingly capable AI systems are robust and beneficial,” the letter states. “Our AI systems must do what we want them to do.” Thousands of people have since signed the letter, including leading artificial intelligence researchers at Google, Facebook, Microsoft and other industry hubs along with top computer scientists, physicists and philosophers around the world. By the end of March, about 300 research groups had applied to pursue new research into “keeping artificial intelligence beneficial” with funds contributed by the letter’s 37th signatory, the inventor-entrepreneur Elon Musk.

Russell, 53, a professor of computer science and founder of the Center for Intelligent Systems at the University of California, Berkeley, has long been contemplating the power and perils of thinking machines.

The machine hums along, quietly scanning the slides, generating Pap smear diagnostics, just the way a college-educated, well-compensated lab technician might.

A robot with emotion-detection software interviews visitors to the United States at the border. In field tests, this eerily named “embodied avatar kiosk” does much better than humans in catching those with invalid documentation. Emotional-processing software has gotten so good that ad companies are looking into “mood-targeted” advertising, and the government of Dubai wants to use it to scan all its closed-circuit TV feeds.

Yes, the machines are getting smarter, and they’re coming for more and more jobs.

To the average video gamer, Pwn Adventure 2’s bear challenge seems impossible. A horde of grizzly bears attacks you from all directions, and no matter how many bears you kill, more always seem to take their place. Even if you amass enough firepower to hold back the onslaught of claws, the dead bears suddenly revive after 90 seconds, pull out AK-47s, and waste your avatar.

Unless, of course, you’re playing Pwn Adventure with a hacker’s mindset. Then you might have reverse-engineered the game’s code and noticed that a bottle of healing wine you drank stored the variable that determined its power locally on your machine, not the game’s server. Change that variable from 20 to 100, and your player is suddenly invincible, ready to massacre as many zombie bears as necessary.

Here’s a mostly useless visualization of the collection of journal articles that sits in my reference database in Endnote. I deal mostly in marine biology, physiology, biomechanics, and climate change papers, with a few molecular/genetics papers thrown in here and there. The database has 3325 entries, 2 of which have ambiguous publication years and aren’t represented above.

… When I think about scientific software as a part of science, I inevitably start with its similarities to building scientific instruments. New instrumentation and methods are absolutely essential to scientific progress, and it is clear that good engineering and methods development skills are incredibly helpful in research.

So, why did the editors at High Profile Journal bounce our paper? I infer that they drew exactly this parallel and thought no further.

But scientific software is only somewhat like new methods or instrumentation.

If you’ve been paying attention to the MongoDB world lately, then you know that exciting things are afoot. The 3.0 release introduces a modular storage engine API, allowing third party engines like RocksDB, TokuDB and WiredTiger to integrate seamlessly with the MongoDB data interface.

The RocksDB engine developed by Facebook engineers is one of the fastest, most compact and write-optimized storage engines available. RocksDB has been running for years as a storage layer for various services here at Facebook, so we have a lot of confidence in its maturity. It is also currently the only LSM engine available for use with MongoDB (WiredTiger also supports LSM natively, but only the B-tree implementation will be available for MongoDB until later releases).

At Parse we have been working closely with other Facebook engineers to integrate and optimize RocksDB + MongoDB, by replaying a wide range of production workloads offline and comparing results between mmapv1, RocksDB, and WT.