November 20, 2018

The Spooky Side of Machine Learning

“When it comes to machine learning, things can get a little spooky: We know an algorithm can solve a problem, but we often don’t know exactly how.

Today’s machine-learning algorithms are considered a form of artificial intelligence, but it’s more helpful to think of them as prediction algorithms: Based on movies that this customer has rated highly, how much do we think he will like this other movie? Based on people we have hired in the past, how likely are we to hire this job candidate? Based on a list of past Halloween costumes, what might humans dress up as this year?

Hoping to find an answer to that last question, we turned to a machine-learning algorithm called textgenrnn that can learn to imitate text. Its author, Max Woolf, designed it as a blank slate ready to learn any kind of text; the text we gave it to imitate was a list of 7,182 costumes that people sent to over a year. Here are some examples of what that algorithm produced.

The algorithm learned to spell all of these words and phrases without human intervention, just by looking at the costume examples we gave it. It starts by making predictions about which letters should be used in which order to make a Halloween costume, and then it checks its own predictions by looking at the data used to train it. If it’s wrong (and at first, it almost always is), it refines its internal structure. Gradually, its predictions get better.”

Read more

JJ Best Of The Web

"...after five months of canceled meetings and muted statements of dissatisfaction by both countries, experts say there is no sign of progress toward the Singapore goal of so-called "denuclearization" of the North."

"The presidential news conferences have become frustrating to watch and, doubtless, are frustrating for President Trump to engage in. While we have freedom of the press in our country, we should not tolerate unprofessionalism."

"It's highly unlikely that Israel's center-left parties will form a coalition to run together in the 2019 election, but they should not abandon efforts to find common ground to fight for."

"Cam has the brilliant audacity to argue that the internet isn’t about connecting people. Netflix’s slick, saucy new horror noir understands the existential terror of losing your carefully curated fictional internet persona."

"Wealthy nations have strong currencies. A strong dollar allows Americans to buy goods, services and resources from other countries at low prices."

"China’s leaders like the internet they have created. And now, they want to direct the nation’s talent and tech acumen toward an even loftier end: building an innovation-driven economy, one that produces world-leading companies."

"At an inaugural desert festival of yogis and spirit guides like Russell Brand, an exclusive industry grapples with consumerism, addiction, and the actual meaning of wellness."

"The confusing thing about Franzen is that even people who hate him admit that he is a great novelist, and even people who love him admit that his essays are often a little on the insufferable..."

"“And just like that” or “in the blink of an eye” are familiar captions on parenting milestone photos. But for me, while the days were long, not even one year flew by."

"How the Silicon Valley set fell in love with sourdough and decided to disrupt the 6,000-year-old craft of making bread, one crumbshot at a time."

"...everyone can — and should — learn quantum mechanics. It’s not rocket science — it’s a fundamental part of how our world works, and not as complex as you might fear."

"New Hebrew University initiative brings international students to Yoga studio run by Breslov Hasidim for course on 'Judaism and the Body.'"