1 result for “dropout”
A regularisation technique in deep learning that randomly deactivates neurons during training, preventing co-adaptation and improving generalisation. Introduced by Hinton and colleagues in 2012 and formalised in 2014.