Exercise 4.6

Answers

If we use the hard-order constraint, with less parameters, the perceptron’s VC dimension decreases, and it’s less likely to classify the same amount of points with more parameters. If we use the soft-order constraint, it won’t change the signs of (wTx) even when w is small. So we’ll still be able to classify the points.

User profile picture
2021-12-08 09:28
Comments