Researchers can officially start submitting algorithms.
A few months ago, Elon Musk and a few other big players in Silicon Valley announced their creation of OpenAI, a non-profit research firm that would provide an open platform for researchers to share and compare algorithms for artificial intelligence.
“Our goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return,” the team wrote in their introductory blog post.
Now, the creators of OpenAI have announced a new platform for the organization, called OpenAI Gym. It’s “a toolkit for developing and comparing reinforcement learning algorithms,” as stated on the website.
Interestingly, OpenAI Gym won’t include leaderboards or competitions based on who can create the top scoring algorithm. One of the biggest challenges in the field of artificial intelligence is achieving “generalization,” so that’s what the initiative will focus on. Generalization means that an artificially intelligent robot would be versatile in completing numerous tasks rather than being proficient at one.
For example, some algorithms can recognize images of animals or faces, while others can recognize speech, but they can’t do each other’s jobs since they approach data in different ways. The goal of generalization is to create algorithms that would know how to deal with both, in the same way that humans can switch from task to task.
"It's not just about maximizing score; it's about finding solutions which will generalize well," clarifies OpenAI Gym's submission documentation. "Solutions which involve task-specific hardcoding or otherwise don't reveal interesting characteristics of learning algorithms are unlikely to pass review."
The idea is that researchers and amateur programmers will build algorithms, and then upload them in various virtual spaces to be tested. Then, they’ll be able to see how their algorithm performed in an objective test and make any necessary adjustments before publishing their work for the rest of the community to critique. You can see the requirements and how to get started on uploading algorithms here.
The OpenAI Gym platform is hinged on reinforcement learning — if an algorithm does well, it receives a reward, and if it fails, it’s given no reward and then tries something different.
“Reinforcement learning (RL) is the subfield of machine learning concerned with decision making and motor control,” as stated on the OpenAI Gym website. “It studies how an agent can learn how to achieve goals in a complex, uncertain environment.”
There will be a number of virtual environments that researchers can test their algorithms in, including Atari games, other board games, simulated 2D and 3D robotics, and more.
OpenAI Gym is now in an open beta mode, so researchers can begin to submit their algorithms. We’re excited to see how creative people will get.
You might also like: Robots Could Learn New Skills by Watching YouTube