Nautilus

The Fundamental Limits of Machine Learning

From a computer’s standpoint, the difficulty in pattern recognition is one of surplus: with an endless variety of patterns, all technically valid, what makes one “right” and another “wrong?”Photograph from Wikicommons

Not long ago, my aunt sent her colleagues an email with the subject, “Math Problem! What is the answer?” It contained a deceptively simple puzzle:

She thought her solution was obvious. Her colleagues, though, were sure their solution was correct—and the two didn’t match. Was the problem with one of their answers, or with the puzzle itself?

My aunt and her colleagues had stumbled across a fundamental problem in machine learning, the study of computers that learn. Almost all of the learning we expect our computers to do—and much of the learning we ourselves do—is about reducing information to underlying patterns, which can then be used

You’re reading a preview, subscribe to read more.

More from Nautilus

Nautilus3 min read
A Crystal Ball for Evolution
Evolution is happening all the time, all around us. It can reshape entire populations of a species in a wink of time—just a couple of generations. Take Darwin’s finches, which are mostly native to the Galapagos Islands. In the 1970s, Peter and Rosema
Nautilus4 min read
Where the Ocean Exhales
When one thinks of Antarctica, one imagines a vast landscape in shades of blinding white, ice and snow stretching as far as the eye can see. But to really consider Antarctica is to consider its water. The Southern Ocean, which encircles Antarctica, i
Nautilus18 min read
The Collapse Is Coming. Will Humanity Adapt?
I’ve known Dan Brooks for 40 years now. Somehow we’re still talking to each other. We’ve followed radically different trajectories since first meeting back in the ’80s. Over the decades, Dan built a truly impressive rap sheet as an evolutionary biolo

Related