Skip to main content
18 events
when toggle format what by license comment
Oct 7, 2019 at 11:05 comment added Neil Slater Not really, but search for "1.7159 tanh" and you will see it is a commonly used variant
Oct 7, 2019 at 10:49 comment added zephyr @NeilSlater - any references you would suggest looking at?
Oct 7, 2019 at 7:50 comment added Neil Slater There are quite a few resources that suggest using $1.7159 \text{tanh}(\frac{2}{3}x)$ . . . I have always assumed that was a similar idea to batch normalisation - about creating closer to ideal distribution of features in hidden layers, including the effect on gradients.
S May 17, 2019 at 7:35 history bounty ended zephyr
S May 17, 2019 at 7:35 history notice removed zephyr
May 15, 2019 at 8:42 vote accept zephyr
May 15, 2019 at 9:34
May 14, 2019 at 16:30 history edited Juan Esteban de la Calle CC BY-SA 4.0
added 1 character in body
May 14, 2019 at 11:01 vote accept zephyr
May 15, 2019 at 8:07
May 14, 2019 at 11:01 answer added zephyr timeline score: 0
May 11, 2019 at 4:47 comment added Tuyen Well, if you use mini-batches - as common now - in training DNN, then probability/statistics/randomness will enter definitely. One easy-to-understand source for this is Nielsen's book Neural Networks and Deep Learning. How about you run several experiments to verify this yourself and report what you found here after?
May 10, 2019 at 19:07 answer added Esmailian timeline score: 11
S May 10, 2019 at 11:13 history bounty started zephyr
S May 10, 2019 at 11:13 history notice added zephyr Authoritative reference needed
May 8, 2019 at 4:02 answer added Tuyen timeline score: 1
May 7, 2019 at 13:43 history edited Green Falcon CC BY-SA 4.0
edited body; edited tags; edited title
May 7, 2019 at 13:40 answer added Green Falcon timeline score: 4
May 7, 2019 at 13:10 review First posts
May 7, 2019 at 18:57
May 7, 2019 at 13:07 history asked zephyr CC BY-SA 4.0