Exercise

From regression to multi-class classification

Recall that the models we have seen for binary classification, multi-class classification and regression have all been similar, barring a few tweaks to the model.

In this exercise, you'll start by building a model for regression, and then tweak the model to perform a multi-class classification.

Instructions 1/2

undefined XP
  • 1
    • Create a neural network with exactly four linear layers, which takes the input tensor as input, and outputs a regression value, using any shapes you like for the hidden layers.
  • 2
    • A similar neural network to the one you just built is provided, containing four linear layers; update this network to perform a multi-class classification with four outputs.