Embedded Files
Project Information
Project Information
Team Size -- 1
Team Size -- 1
Software & Languages -- Visual Studio; JavaScript; HTML
Software & Languages -- Visual Studio; JavaScript; HTML
ML Package -- Node.js
ML Package -- Node.js
Client -- Nokia Technology Leadership Council
Client -- Nokia Technology Leadership Council
Project Summary
Project Summary
How do you introduce something as complicated and in-depth as Machine Learning (ML) to elementary school children? They don't know what a neural network is, nor do they know how it works. They probably don't even know how to code. But, they do know how to sort things, and that was the project Nokia Technology Leadership Council requested: Build a sorting game that implemented an ML predictor in the background.
How do you introduce something as complicated and in-depth as Machine Learning (ML) to elementary school children? They don't know what a neural network is, nor do they know how it works. They probably don't even know how to code. But, they do know how to sort things, and that was the project Nokia Technology Leadership Council requested: Build a sorting game that implemented an ML predictor in the background.
Developer Commentary
Developer Commentary
The ML Game was demonstrated as part of Nokia of Naperville, IL's STEM outreach programs. I presented the game to both elementary-aged children and their parents at multiple open houses, describing the basics of how it worked.
The ML Game was demonstrated as part of Nokia of Naperville, IL's STEM outreach programs. I presented the game to both elementary-aged children and their parents at multiple open houses, describing the basics of how it worked.
When presented to younger audiences, I limited what it could generate and which sorting area it had. Only four colors were used, and size was eliminated to reduce complexity.
When presented to younger audiences, I limited what it could generate and which sorting area it had. Only four colors were used, and size was eliminated to reduce complexity.
Once enough shapes were sorted, the ML algorithm would attempt to predict where any unsorted shapes should go. The location would be represented by a little dot inside a miniature sorting area shown within each object. The confidence of the output would be represented by its transparency, with faded dots representing less confident predictions.
Once enough shapes were sorted, the ML algorithm would attempt to predict where any unsorted shapes should go. The location would be represented by a little dot inside a miniature sorting area shown within each object. The confidence of the output would be represented by its transparency, with faded dots representing less confident predictions.
A more complicated and impressive version included differences in object size and randomized colors within the color wheel. The sorting area also became a bullseye target, with predictions determining where in the target the objects should go.
A more complicated and impressive version included differences in object size and randomized colors within the color wheel. The sorting area also became a bullseye target, with predictions determining where in the target the objects should go.
The controls for what parameters got used to randomly generate the shapes were in the options menu. The menu also included the controls for what type of sorting area was shown.
The controls for what parameters got used to randomly generate the shapes were in the options menu. The menu also included the controls for what type of sorting area was shown.
Page updated
Google Sites
Report abuse