A good thing to share for now :- Artificial Intelligence neural networks !

  • 3
    What concept do I google to understand this?
  • 6

    You can find on official site of Google Learning

    BTW, there are many tutorial available on net..
    But this is just official from Google.
  • 7
    See, people? A bit more than a bunch of if statements.
  • 3
  • 1
    Not that helpful unless you have more than a guess about what each type of cell does :/ really interesting though!
  • 0
    Maybe DNN too?
  • 0
    Bring on the flood of “If statement” comments and jokes circle jerk begin!
  • 3
    Meh... There's no such thing as a "Deep Feed Forward". What they call a "feed forward" is just a multi layer perception (multi layer = deep)... Feed forward is one of the main algorithms that is used in (probably) every type of neural network. Kinda BS this chart; neural network architectures can't be simplified to those nice little diagrams. A multi layer perception could very well have the same topology as a CNN, but they differ in the implementation.
    That was all, good night.
  • 0
    @transpalette I agree, but these are standard topologies that you'll find in use pretty much everywhere. Of course you can have a MLP with the same arch as a CNN but then you just might as well call it a CNN instead.

    Also, the deep feedforward is just a feedforward with more hidden layers, lol. That's all. Researchers frequently differentiate between deep and shallow networks. And yes, the feedforward bit is standard across most networks, but then again, that's just a naming issue. Again, these are all standard networks which are in (commercial) use.
  • 1
    @RememberMe yeah that's what I'm saying about deep neural nets, they're just made of hidden layers. But no, an MLP with the same topology as a CNN doesn't become a CNN ;) CNNs aren't made of perceptrons (except for the fully connected part) but of different types of neurons (filters to be precise).
Add Comment