But just keep in mind that neural networks beyond 4 layers are often not better. and most work can be done with 3 layers total.. (so that is one hidden one), having two hidden ones can just do some things that a 3layer cant.
But more then 4 is more often a bad design and usually overkill of layers for most practical usages.
Although theoretical there no math limit to the amount of layers in practical terms its more often calculation time that explodes to unless its a simple FANN based network
but then still 3 or 4 layers is 95% of the time enough
I've heard this many times. Not contradicting it (I believe it, for the tasks we tend to give ANNs, anyways) Do you have a good review/citation for this?
Yesterday i had to explain to some people i work with what Microsoft is, they dont know it, and dont understand it, they are used to android these days. And find the whole thing confusing why pay for it ?.. its hard to justify or explain..
i am planning for a DIY mostly because of i want to be able to produce something fast (not waiting on delivery), and because i like the engineering part.
It is so simple, this is the same object seen over 2 places.
Think of a fishing tank, the side glass wall and the front glass wall show the same fish. And that's i hope an easy to understand example of an extra dimension, it solves many physics problems add a dimension..
As the fish swims in the tank, both walls show change in position to and no need for communication at all, its just one fish.
you need proof for extra-dimensional existence just take all those experiments that could be solved by it .. double split... spooky action .... and many more.. (even quantum computers inner working)
I also use spead reading once in a while.
I can read realy fast going through large documents
To be able to find something quickly
I once got a huge document on my desk 3400 pages, and i had to decide if we would do that project, please try to read it.
To their surprice i fineshed reading and pointed them on the hard parts of it, they never thought anyone would be able to read the entire doc in just a few hours for the deadline.
And there was no link to the other meteor that passed earth today ??. Or did Nasa coverup that it didnt knew what trajectory it would take. It seams unlikely to me that those two events are not related.
I mean recently Nasa found some anomaly on Mars too, and then they say "no its not interesting its just erosion; (mars has next to none atmospheric pressure btw) i mean if i where a geologist, then any curious shape even if it was made by erosion would attract my attention i would take samples; and most likely such samples would be more interesting then dust-hole digging. What has become from NASA i wonder. Once they where explorers curious.. now they only call their device curiosity
But more then 4 is more often a bad design and usually overkill of layers for most practical usages.
Although theoretical there no math limit to the amount of layers in practical terms its more often calculation time that explodes to unless its a simple FANN based network but then still 3 or 4 layers is 95% of the time enough