I asked the boys what they wanted to do for our math project today and the both wanted to learn more about neural networks. Our project from yesterday is here:
and the program we are using is here:
The thing the boys were most interested in were the drop down features at the top of the program. Essentially their questions were:
(1) What does the activation feature do?
(2) What does regularization mean?
(3) What is the difference between classification and regression?
So . . . all of these questions are a little bit beyond 5th and 7th grade, but I did my best (though I punted on (2) to try to give a better explanation of the others).
In the first part we talked about the difference between classification and regression:
In the second part we talked about what activation functions do. This idea is probably well beyond what kids can understand in any detail, but the program actually proved to be a good tool to illustrate the point.
It is actually amazing to hear what kids have to say when they are trying to digest some of these ideas.
So, I let them play around with the program and investigate which activation functions worked well / not well with different data sets. Again, the ideas here are difficult for kids to grasp, but they did a pretty good job thanks to the help from the program.
So, a fun couple of days playing around with some simple neural network ideas with the boys. Not sure what I’m going to do if they want to learn more – ha ha!