Friday, January 6, 2023

The Physics Principle That Inspired Modern AI Art

Math and Science News from Quanta Magazine
View this email in your browser
My Bookmarks

MACHINE LEARNING | ALL TOPICS

 

The Physics Principle That Inspired Modern AI Art

By ANIL ANANTHASWAMY

Diffusion models generate incredible images by learning to reverse the process that, among other things, causes ink to spread through water.

Read the article

COMBINATORICS

 

Google Researcher, Long Out of Math, Cracks Problem About Sets

By KEVIN HARTNETT

On nights and weekends, Justin Gilmer attacked an old question in pure math using the tools of information theory.

Read the article


Related: 
How Shannon Entropy Imposes
Fundamental Limits on Communication

By Kevin Hartnett

EXPLAINERS

 

Inside Ancient Asteroids, Gamma Rays Made Building Blocks of Life

By JOHN RENNIE & ALLISON PARSHALL

A new radiation-based mechanism adds to the ways that amino acids could have been made in space and brought to Earth.

Read the article


Related: 
Peptides on Stardust May
Have Provided a Shortcut to Life

By Yasemin Saplakoglu

QUANTA SCIENCE PODCAST

 

A Good Memory or a Bad One? One Brain Molecule Decides.

Podcast hosted by SUSAN VALOT;
Story by YASEMIN SAPLAKOGLU

When the brain encodes memories as positive or negative, one molecule determines which way they will go.

Listen to the podcast

Read the article

Around the Web

The Sound of Stars
Astronomers are translating their data into sounds to make their field more accessible to blind and visually impaired people, writes Timmy Broderick for Scientific American. "Seeing" the data in new ways could potentially lead to new discoveries too. In 2017, astronomers applied their musical knowledge to the TRAPPIST-1 system and found that the orbital periods of its exoplanets formed a major ninth chord. For Quanta, Joshua Sokol wrote about the "music of the spheres" that the scientists composed based on this data.


Nice Signature
A new kind of "Bayesian machine" can learn to recognize a person's signature with less training than a standard neural network, reports Nature Magazine. The microprocessor uses Bayesian reasoning to draw conclusions from incomplete information. Computer scientist Judea Pearl, who conceived of Bayesian networks in the 1980s, is now teaching computers about cause and effect. In 2018 for Quanta, Kevin Hartnett interviewed Pearl about this key step to building a truly intelligent machine.
Follow Quanta
Simons Foundation

160 5th Avenue, 7th Floor
New York, NY 10010

Copyright © 2023 Quanta Magazine, an editorially independent division of the Simons Foundation

Scientist Pankaj

Today in Science: How to stop an apocalyptic asteroid strike

...