Last Week in AI

Source: Deep Learning on Medium

Every week, my team at Invector Labs publishes a newsletter that covers the most recent developments in AI research and technology. You can find this week’s issue below. You can sign up for it below. Please do so, our guys worked really hard on this:

From the Editor: Solving the Rubik’s Cube Like a Human Would

Building artificial intelligence(AI) systems that develop knowledge and reason through tasks in the same way humans do its one of the pivotal challenges of the next decade of AI. While many AI systems have shown impressive results across diverse cognitive tasks, their way of developing knowledge still remains fairly mechanic. Solving the Rubik’s cube is one of those tasks that have been solved many times by AI but in ways that don’t come even closer to human cognitive processes.

This week, AI powerhouse OpenAI astonished the world by unveiling a prototype of a robotic arm that could solve a Rubik’s cube with one robotic hand. The technique used a combination of reinforcement learning and a new technique that introduced constant challenges into the learning process to maximize the learning abilities of the algorithm. OpenAI also leveraged a robotic arm that showed impressive levels of dexterity and abilities to react to unforeseen environmental challenges. While the OpenAI demonstration was constrained to the Rubik’s cube, the principles can be applied to build a new generation of AI systems that develop knowledge following the similar processes to humans. Just a bit faster 😉

Now let’s take a look at the core developments in AI research and technology this week:

AI Research

The Robot that can Solve the Rubik’s Cube With One Hand

In an impressive display of both robotics and AI skills, OpenAI published a research of a system that can solve Rubik’s cube reasoning like a human would.

>Read more in this blog post from OpenAI

Billion-Scale Image Classification

Facebook AI Research(FAIR) published a paper that details a technique that combines semi and weakly supervised learning for massively scalable image classification models.

>Read more in this blog post from the FAIR team

MIT Fights Fake News

MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) published two papers proposing new techniques to fight fake news.

>Read more in this blog post from MIT News

Cool AI Tech Releases

A Framework for Building Reasoning Systems

Facebook open sourced ReAgent, a new framework for designing systems that can make and rely on decisions.

>Read more in this blog post from the Facebook engineering team

Restoring Ancient Greek Texts Using AI

DeepMind open sourced PYTHIA, the first ancient text restoration model that recovers missing characters from a damaged text input using neural networks.

>Read more in this blog post from the DeepMind team

Uber’s Michelangelo Improvements

Uber published a series of updates to its Michelangelo system which is the backbone of machine learning at the transportation powerhouse.

>Read more in this blog post from the Uber engineering team

AI in the Real World

Data-labelling Startups are Fueling AI Adoption

The most boring part of AI might become a $5 billion industry in a few years.

>Read more in this coverage from The Economist

AI Powered Chess Board

Square Off launched a Kickstarter campaign to build an intelligent chess board that uses AI to improve the player’s training.

>Read more in this coverage from VentureBeat

America Needs More than Silicon Valley to Compete with China

The New York Times published an interesting piece about the America’s reliance on a handful of technology giants to compete with China’s big state-funded AI initiatives.

>Read more in this article from The New York Times