Mikael Laine
Mikael Laine
  • Видео 4
  • Просмотров 118 607
Neural Network Backpropagation Example With Activation Function
The simplest possible back propagation example done with the sigmoid activation function.
Some brief comments on how gradients are calculated in actual implementations.
Edit: there is a slight omission/error in the da/dw expression, as pointed out by Laurie Linnett. The video has da/dw = a(1-a), but it should be ia(1-a), because the argument to a is the function (iw), whose derivative (with respect to w) is i.
Просмотров: 36 514

Видео

Another Simple Neural Network Backpropagated
Просмотров 6 тыс.6 лет назад
The second simplest neural networks explained mathematically: how to make sense of the gradients and optimization in detail for a very simple neural network. Please refer to the first video if you haven't seen it and want more: ruclips.net/video/8d6jf7s6_Qs/видео.html This is part 2/2 of the "Simple Neural Networks Backpropagated Series"
Implementation of the Absolutely Simplest Neural Network
Просмотров 2,4 тыс.6 лет назад
This is provided for completeness' sake: the code for the neural network in the previous video. Code here: github.com/mikaellaine/random/blob/master/simplest.py Install Tensorflow as described here: www.tensorflow.org/install/
0:03 / 9:21The Absolutely Simplest Neural Network Backpropagation Example
Просмотров 74 тыс.6 лет назад
Easy explanation for how backpropagation is done. Topics covered: - gradient descent - exploding gradients - learning rate - backpropagation - cost functions - optimization steps

Комментарии

  • @jeremygarrard6311
    @jeremygarrard6311 4 дня назад

    Amazing explanation!! Any chance you can add in a bias and show how that works too?

  • @AlejandroFernandezDaCosta
    @AlejandroFernandezDaCosta 5 дней назад

    very good explanation

  • @mrinky8129
    @mrinky8129 17 дней назад

    amazing explanation

  • @fengyqy
    @fengyqy Месяц назад

    Thanks for the video!!! at 11:09, da/dw = a*(1-a), hmmm.. as it is e^-(iw), so should it be i*a*(1-a)? or in other words, w' = w - r*i*(da/dw)*(dC/da). Thanks again

  • @hasen_judi
    @hasen_judi 3 месяца назад

    I've been trying to find a way to learn back propagation without understanding deep math. Probably not a good idea lol. But my purpose is to implement a neural network with back propagation by hand. The comment at 13:00 really helped elucidate the path forward for me. I think I still don't have a deep understanding of the math, but at least I now have a path for in-code implementation. Thanks for making this video.

  • @JoeBurnett
    @JoeBurnett 4 месяца назад

    Fantastic video! I wish you were still making videos on the subject of AI with this teaching method.

  • @wilfredomartel7781
    @wilfredomartel7781 5 месяцев назад

    great video!

  • @ingoampt
    @ingoampt 6 месяцев назад

    What about when we have like activation function like relu. Or etc ?

  • @gulamm1
    @gulamm1 7 месяцев назад

    The best explanation.

  • @jancsi-vera
    @jancsi-vera 8 месяцев назад

    Wow, thank you

  • @3r1kz
    @3r1kz 8 месяцев назад

    I don't know anything about this subject but I was understanding it until the rate of change function. Probably a stupid question but why is there a 2 in the rate of change function, as in 2(a-y). Is this 2 * (1.2 - 05)? Why the 2? I can't really see the reference to the y = x^2 but that's probably just me not understanding the basics. Maybe somebody can explain for a dummy like me. Wait maybe I understand my mistake, the result should be 0.4 right? So its actually 2(a-1) because otherwise multiplication goes first and you end up with 1.4?

    • @joemurray1
      @joemurray1 7 месяцев назад

      The derivative of x^2 (x squared) is 2x. The cost function C is the square of the difference between actual and desired output i.e. (a-y)^2. Its derivative (slope) with respect to a is 2(a-y). We don't use the actual cost to make the adjustment, but the slope of the cost. That always points 'downhill' to zero cost.

  • @cachaceirosdohawai3070
    @cachaceirosdohawai3070 9 месяцев назад

    Any help dealing with multi-neuron layers?, the formulas in 11:19 look different for multi-neuron layers

    • @mikaellaine9490
      @mikaellaine9490 9 месяцев назад

      Check my channel for another example with multiple layers.

  • @thamburus7332
    @thamburus7332 9 месяцев назад

  • @АртурЗарипов-б2й
    @АртурЗарипов-б2й 9 месяцев назад

    Good job! Thank you very much!

  • @ss5380
    @ss5380 10 месяцев назад

    You are a life saver!! Thank you for breaking the whole process down in such an understandable way!!

  • @vudathabhavishya9629
    @vudathabhavishya9629 11 месяцев назад

    Can anyone explain how to plot for 2(a-y),c=(a-y)2. i=1.5

  • @rafaelramos6320
    @rafaelramos6320 Год назад

    Hi, a = i * w 1.5. * 2(a -y) = 4.5 * w - 1.5 What happened to the y?

    • @LaurentPrat
      @LaurentPrat 10 месяцев назад

      y is given = the target value, here = 0.5. => 1.5*2(1.2-0.5) = 2.1 which equal to 4.5*0.8-1.5

  • @semtex6412
    @semtex6412 Год назад

    on 2:40, Mikael mentioned "...and the error therefore, is 0.5" i think he meant "and the *desired output*, therefore is 0.5"? slight erratum perhaps?

    • @semtex6412
      @semtex6412 Год назад

      because otherwise, the cost (C) is 0.49, not 0.5

  • @raaziyahshamim4761
    @raaziyahshamim4761 Год назад

    What software did you use to write the stuff.. good lecture

  • @talhanaeemrao4305
    @talhanaeemrao4305 Год назад

    There are some videos which you wish that it never end. This video in among top of these.