7 comments

[ 4.2 ms ] story [ 27.1 ms ] thread
I'd imagine it's a lot easier and more effective to teach middle schools basic probability first. All you need is a coin or a deck of cards to introduce them to real world examples.
Not to mention that machine learning without at least some probabilistic intuition is asking for trouble in many applications.
for image classification? I don't think so.
Maybe for copying and pasting MNIST-in-Keras.

But to even understand and make good use of cross-validation, you absolutely need to understand expected values, sample size, and the bias/variance tradeoff. That's all statistical intuition that depends more or less directly on a basic understanding of and healthy respect for probability.

(comment deleted)