Self-Motivation
David Goggins
Former Navy Seal
Career Development
Bryan Cranston
Actor
Critical Thinking
Liv Boeree
International Poker Champion
Emotional Intelligence
Amaryllis Fox
Former CIA Clandestine Operative
Management
Chris Hadfield
Retired Canadian Astronaut & Author
Learn
from the world's big
thinkers
Start Learning

New AI improves itself through Darwinian-style evolution

AutoML-Zero is a proof-of-concept project that suggests the future of machine learning may be machine-created algorithms.

New AI improves itself through Darwinian-style evolution
Pixabay
  • Automatic machine learning is a fast-developing branch of deep learning.
  • It seeks to vastly reduce the amount of human input and energy needed to apply machine learning to real-world problems.
  • AutoML-Zero, developed by scientists at Google, serves as a simple proof-of-concept that shows how this kind of technology might someday be scaled up and applied to more complex problems.

Machine learning has fundamentally changed how we engage with technology. Today, it's able to curate social media feeds, recognize complex images, drive cars down the interstate, and even diagnose medical conditions, to name a few tasks.

But while machine learning technology can do some things automatically, it still requires a lot of input from human engineers to set it up, and point it in the right direction. Inevitably, that means human biases and limitations are baked into the technology.

So, what if scientists could minimize their influence on the process by creating a system that generates its own machine-learning algorithms? Could it discover new solutions that humans never considered?

To answer these questions, a team of computer scientists at Google developed a project called AutoML-Zero, which is described in a preprint paper published on arXiv.

"Human-designed components bias the search results in favor of human-designed algorithms, possibly reducing the innovation potential of AutoML," the paper states. "Innovation is also limited by having fewer options: you cannot discover what you cannot search for."

Automatic machine learning (AutoML) is a fast-growing area of deep learning. In simple terms, AutoML seeks to automate the end-to-end process of applying machine learning to real-world problems. Unlike other machine-learning techniques, AutoML requires relatively little human effort, which means companies might soon be able to utilize it without having to hire a team of data scientists.

AutoML-Zero is unique because it uses simple mathematical concepts to generate algorithms "from scratch," as the paper states. Then, it selects the best ones, and mutates them through a process that's similar to Darwinian evolution.

AutoML-Zero first randomly generates 100 candidate algorithms, each of which then performs a task, like recognizing an image. The performance of these algorithms is compared to hand-designed algorithms. AutoML-Zero then selects the top-performing algorithm to be the "parent."

"This parent is then copied and mutated to produce a child algorithm that is added to the population, while the oldest algorithm in the population is removed," the paper states.

The system can create thousands of populations at once, which are mutated through random procedures. Over enough cycles, these self-generated algorithms get better at performing tasks.

"The nice thing about this kind of AI is that it can be left to its own devices without any pre-defined parameters, and is able to plug away 24/7 working on developing new algorithms," Ray Walsh, a computer expert and digital researcher at ProPrivacy, told Newsweek.

If computer scientists can scale up this kind of automated machine-learning to complete more complex tasks, it could usher in a new era of machine learning where systems are designed by machines instead of humans. This would likely make it much cheaper to reap the benefits of deep learning, while also leading to novel solutions to real-world problems.

Still, the recent paper was a small-scale proof of concept, and the researchers note that much more research is needed.

"Starting from empty component functions and using only basic mathematical operations, we evolved linear regressors, neural networks, gradient descent... multiplicative interactions. These results are promising, but there is still much work to be done," the scientists' preprint paper noted.

Take your career to the next level by raising your EQ

Emotional intelligence is a skill sought by many employers. Here's how to raise yours.

Gear
  • Daniel Goleman's 1995 book Emotional Intelligence catapulted the term into widespread use in the business world.
  • One study found that EQ (emotional intelligence) is the top predictor of performance and accounts for 58% of success across all job types.
  • EQ has been found to increase annual pay by around $29,000 and be present in 90% of top performers.
Keep reading Show less

Just How Much Land Does the Federal Government Own — and Why?

The rough beauty of the American West seems as far as you can get from the polished corridors of power in Washington DC.

Surprising Science

The rough beauty of the American West seems as far as you can get from the polished corridors of power in Washington DC. Until you look at the title to the land. The federal government owns large tracts of the western states: from a low of 29.9% in Montana, already more than the national average, up to a whopping 84.5% in Nevada.

Keep reading Show less

Break down the secrets to a lucrative project management career

This course collection can get you trained and ready for a six-figure career in this field.

Gear
  • The Premium 2020 Project & Quality Management Certification Bundle explores the most popular project management methodologies.
  • Coursework covers Agile, Agile Scrum, PMI-PMBOK and Six Sigma approaches.
  • Valued at $2,699, the course package is on sale for just $45.99.
Keep reading Show less

Can VR help us understand layers of oppression?

Researchers are using technology to make visual the complex concepts of racism, as well as its political and social consequences.

Future of Learning
  • Often thought of first as gaming tech, virtual reality has been increasingly used in research as a tool for mimicking real-life scenarios and experiences in a safe and controlled environment.
  • Focusing on issues of oppression and the ripple affect it has throughout America's political, educational, and social systems, Dr. Courtney D. Cogburn of Columbia University School of Social Work and her team developed a VR experience that gives users the opportunity to "walk a mile" in the shoes of a black man as he faces racism at three stages in his life: as a child, during adolescence, and as an adult.
  • Cogburn says that the goal is to show how these "interwoven oppressions" continue to shape the world beyond our individual experiences. "I think the most important and powerful human superpower is critical consciousness," she says. "And that is the ability to think, be aware and think critically about the world and people around you...it's not so much about the interpersonal 'Do I feel bad, do I like you?'—it's more 'Do I see the world as it is? Am I thinking critically about it and engaging it?'"
Keep reading Show less
Coronavirus

Russia claims world's first COVID-19 vaccine but skepticism abounds

President Vladimir Putin announces approval of Russia's coronavirus vaccine but scientists warn it may be unsafe.

Scroll down to load more…
Quantcast