While the jury is still out if Artificial Intelligence (AI) is going to be a good thing or bad thing for humanity, one thing that we have all wondered is how exactly does the whole thing work? If you’re one of those who has always been curious about understanding the way machine learning works, Google might have something interesting to offer.
Introducing Google’s in-browser experiment called Teachable Machine. Being considered a simple two-minute summary of what a lot of modern AI can or can’t do, Google’s new browser experiment lets one use their webcam to train an extremely basic AI program.
In order to use Teachable Machine, a user just has to press the “train green/purple/orange” buttons so that the machine records whatever it can see through the webcam. Once the AI has learned enough, it will provide the user the output in whichever way the user likes (a sound effect or a GIF or even some speech) when it sees the activity or object the user has trained it with.
This experiment lets anyone explore how machine learning works, in a fun, hands-on way. One can teach a machine to use their camera, live in the browser– with absolutely no coding required whatsoever. It basically involves training a neural network locally on your device, without sending any images to a server. That’s how it responds so quickly.
The image recognition is powered by a neural network. It was made possible by Nikhil Thorat, part of the team behind deeplearn.js. It’s an open-source library that allows web developers to train and run machine learning models locally in the browser. The code for this experiment is open-sourced on Github.
The experiment demonstrates some fundamental aspects of machine learning. Firstly, that programs like these learn by example. They look to find patterns, and then they remember them. Secondly, they need a lot of examples to learn from. And thirdly, their understanding of the world is basically superficial.
This experiment is a collaborative effort by friends from Støj, a Copenhagen-based creative coding studio; Use All Five, a Los Angeles-based design studio and Creative Lab and PAIR teams at Google.
This development was first reported in The Verge.
[Image: YouTube]
Introducing Google’s in-browser experiment called Teachable Machine. Being considered a simple two-minute summary of what a lot of modern AI can or can’t do, Google’s new browser experiment lets one use their webcam to train an extremely basic AI program.
In order to use Teachable Machine, a user just has to press the “train green/purple/orange” buttons so that the machine records whatever it can see through the webcam. Once the AI has learned enough, it will provide the user the output in whichever way the user likes (a sound effect or a GIF or even some speech) when it sees the activity or object the user has trained it with.
This experiment lets anyone explore how machine learning works, in a fun, hands-on way. One can teach a machine to use their camera, live in the browser– with absolutely no coding required whatsoever. It basically involves training a neural network locally on your device, without sending any images to a server. That’s how it responds so quickly.
The image recognition is powered by a neural network. It was made possible by Nikhil Thorat, part of the team behind deeplearn.js. It’s an open-source library that allows web developers to train and run machine learning models locally in the browser. The code for this experiment is open-sourced on Github.
The experiment demonstrates some fundamental aspects of machine learning. Firstly, that programs like these learn by example. They look to find patterns, and then they remember them. Secondly, they need a lot of examples to learn from. And thirdly, their understanding of the world is basically superficial.
This experiment is a collaborative effort by friends from Støj, a Copenhagen-based creative coding studio; Use All Five, a Los Angeles-based design studio and Creative Lab and PAIR teams at Google.
This development was first reported in The Verge.
[Image: YouTube]
Advertisements