[mlpack-git] [mlpack] Create a function to visualize the features learned by sparse autoencoder (#465)
stereomatchingkiss
notifications at github.com
Wed Oct 28 04:25:27 EDT 2015
>if we train an autoencoder with a single hidden unit on the same input over and over again, then wouldn't the input that maximally activates that single hidden unit be the same input?
Do not know what do you mean by this. Could you describe it further?Base on your description
arma::mat trainData;
for(size_t i = 0; i != 10; ++i)
{
SparseAutoencoder<> sa(....);
//after training
MaximalInputs(sa.Parameters(), output);
//omit trouble some begin, end
bool isSame = std::equal(std::begin(output), std::end(output), std::begin(outputBuffer));
BOOST_REQUIRE(is_same == true);
}
But is this test make sense?
---
Reply to this email directly or view it on GitHub:
https://github.com/mlpack/mlpack/pull/465#issuecomment-151761289
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.cc.gatech.edu/pipermail/mlpack-git/attachments/20151028/54ef7364/attachment.html>
More information about the mlpack-git
mailing list