7
$\begingroup$

I've been thinking if it would be possible to teach something about the basic mathematics behind Neural Networks / Deep learning to a class of 6th graders (12-13ys old). A 90-minute (max) lecture must be assumed.

I know that it maybe sounds like too difficult/impossible -- but I'd love to hear if there are any good ideas. We can assume that the students can each have their personal tablet in-class, and also there is a projector in-class, if that helps in any way.

The goal, assuming that a valid solution does exist, would be showing students one more cool application of maths; especially one that is everywhere those days.

Improve this question
$\endgroup$
3
  • 2
    $\begingroup$ this sounds very interesting, but IMHO 6th grade is too young for the mathematics, but it may still be appropriate to discuss the concepts and general theory without any specifics. That being said, I have no clue how you would go about doing this.. $\endgroup$
    – celeriko
    Apr 1 '16 at 12:22
  • $\begingroup$ How competent are the 6th graders? Gifted, struggling or somewhere in between? Or does the class have a spectrum. $\endgroup$
    – Amy B
    Apr 2 '16 at 17:24
  • $\begingroup$ Although "deep learning" is a buzzword-compliant rebranding of "neural networks", it's not clear to me that we would want to teach young kids that dubious rebranding is "science"... "Neural networks" is already hype enough, I think. $\endgroup$
    – paul garrett
    Apr 3 '16 at 22:52
1
$\begingroup$

Neural networks are largely about doing a large number of regression problems --- sometimes linear regression --- so you may be able to appeal to some intuitive sense about what a best fit line/plane and go from there. It feels like it would be tough to get students to understand how the network learns (back propagation) in a meaningful way without some notion of gradient descent. (Aren't parabolas introduced in later grades?)

However, if what you want to do is show them a modern data analysis tool, I would suggest k-means clustering instead. https://en.wikipedia.org/wiki/K-means_clustering The most advanced concepts used there are distance and averaging (the mean).

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.