0
$\begingroup$

The following images are

a) The weights of a logistic regression model trained on MNIST.

b) The sign of the weights of a logistic regression

How do these images represent the weights?

Would be grateful for any help.

Source of the research paper

enter image description here

$\endgroup$
4
  • 1
    $\begingroup$ It depends on what you are training the model for. If you are trying to classify a face weights will be given to things like eyes nose etc. Just think how would you recognise something, and thats how the LR model will distribute the weights, how the features affect the classification (this is an overgenralization, but you get the idea) $\endgroup$ – user9947 Dec 31 '20 at 15:43
  • $\begingroup$ I understand, maybe you can write it as an answer $\endgroup$ – Hrushi Dec 31 '20 at 17:00
  • 1
    $\begingroup$ Sorry it takes a long time to write an answer, and I am very short of time. But this actually a very common concept. You can find this in any introductory NN book. This is basically what we call Hebbian Learning. The very first paradigm of computational learning. $\endgroup$ – user9947 Dec 31 '20 at 17:02
  • $\begingroup$ Thank you very much! $\endgroup$ – Hrushi Dec 31 '20 at 17:19

Your Answer

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

Browse other questions tagged or ask your own question.