Regularization by Axiliary Output in Deep Learning

[soft talon]

I've been reading a book and they mention a use case for multiple outputs in a DL model that I don't really understand how it works.

There are also many use cases in which you may want to have multiple outputs:
...

• Another use case is as a regularization technique (i.e., a training constraint
  whose objective is to reduce overfitting and thus improve the model’s ability to
  generalize). For example, you may want to add an auxiliary output in a neural
  network architecture (see Figure 10-15) to ensure that the underlying part of the network learns something useful on its own, without relying on the rest of the
  network.

Would anyone happen to know how the auxiliary output works in regularizing the model and could explain it to me?

@edgy oxide Would you happen to know?

[misty flare]

Curious what this is when you figure it out! Seems interesting

[soft talon]

I don't think anyone will be able to help me 😭

[misty flare]

Yeah I have no clue. The diagram made sense other then input wide skipping the hidden layers and going str8 to Concat

[tacit summit]

@soft talon How much background knowledge do you have / what kind of familiarity with ML stuff can I assume?

[misty flare]

Wang in school for data science iirc

[soft talon]

^^

[tacit summit]

Imma just post this here and hope that clears things up, let me know if the post doesn't make any sense: https://stackoverflow.com/questions/43216513/significance-of-auxiliary-output-in-multi-input-and-multi-output-model-using-dee