Subscribe

Sep 13 2016

Machine Learning, Math, and Magic

Issue #39

Do you need a math degree to dig deep into machine learning? Not according to our Article of the Week, which is about how you need less than a year to become a ML professional - and not according to us! We're starting a blog series on taking the math and magic out of machine learning, and you can read the first instalment here.

In the news this week, Google's DeepMind is learning how to talk, and if it didn't take a processor the size of a power plant (that's an exaggeration), it would be pretty record breaking stuff.

Plus, everyone is talking the Stanford paper, pixilation is a thing of the past, and AI has failed to make the internet smarter.

Not signed up yet? Get your weekly dose of superintelligence at Leviathan.ai

Have feedback? We'd love to hear from you. We are @LeviathanAI on Twitter or you can always just reply to this email.

News From the Deep
Google's DeepMind Artificial Intelligence Has Figured Out How to Talk

Google's DeepMind Artificial Intelligence Has Figured Out How to Talk

They've broken an important barrier, but we're still being held back by computational power.

While WaveNet sounds more like a human voice than existing artificial voice generators — known as "text-to-speech" (TTS) systems — it requires too much computing power to make it practical.
The Problem With the Stanford Report’s Sanguine Estimate on Artificial Intelligence

The Problem With the Stanford Report’s Sanguine Estimate on Artificial Intelligence

Everyone is talking Stanford, from whether AI will be predicting crime to why Stanford thinks there will be so little of it.

Why the light treatment of security, barely more than seven paragraphs in an almost four-dozen page report?
Machine Learning Can Identify Pixelated Faces

Machine Learning Can Identify Pixelated Faces

Bad news for privacy, as we teach machines to see like us, and they get better at it than we are.

...software that can undermine the privacy benefits of standard content-masking techniques like blurring and pixelation by learning to read or see what’s meant to be hidden in images.
Industry Responds to Artificial Intelligence Technology Development

Industry Responds to Artificial Intelligence Technology Development

Why can't we seem to answer the question, will robots destroy the world?

Responses to a White House request for information about the future of artificial intelligence show a continued divide.
Why Hasn't Artificial Intelligence Made the Internet Smarter?

Why Hasn't Artificial Intelligence Made the Internet Smarter?

A neat look at why search is still so bad, when AI has gotten so very good.

Will artificial intelligence (AI) make the Internet smarter? Probably someday, but don’t look for it in the foreseeable future.

Article of the Week
Machine Learning in a Year – Learning New Stuff

Machine Learning in a Year – Learning New Stuff

This guy taught himself to be a ML expert in a year - okay, well, a ML professional, at any rate. Great read.

The truth is you don’t need much maths to get started with machine learning, and you don’t need a degree to use it professionally.

Quicklinks

When the headline is all the news you need.

“By far the greatest danger of Artificial Intelligence is that people conclude too early that they understand it.” ― Eliezer Yudkowsky
Archive