I 've never seen such a perfect explanation.I have faced with a problem while I was writing my kernel and searched everywhere to find a solution for understanding reason of error I ve got.Suddenly I saw the video and started to listen carefully .Thank you so much sir
Part 1 and Part 2 are the best explanation of these subjects going. Thanks so much Bevan. You have a real talent for explaining a difficult subject in a way that’s as easy as possible to grasp. Brilliant videos
Simply the best explanations i have found if i could give part one and two a hundred likes i would...Please do more AI content you have a gift for teaching.
Dear Bevan, You have presented both the forward and backward propagation concepts in an exceptional manner, especially the example that you have chosen for doing so. The majority of ANN tutorials stick to 'cat-dog' kind of categorical examples. I have one query though. Do you recommend the use of a non-linear activation function in the output layer? And if yes, which non-linear activation function will you prefer for continuous problems. The thing to keep in mind while choosing this output layer activation function is that it should be able to provide an output which can exceed 1 or can be less than 0 or -1. Sigmoid or tanh activation function in the output layer will not allow us to have such 'greater than 1 or less than 0 / -1' outputs. I hope to hear from you. Thanks!
This is a fantastic explanation, but you'll need some knowledge of machine learning from scratch, particularly linear and logistic regression. This is perfect and exactly what I needed to understand NN. Liked and subbed.
Hi Lucia. Yes it was a random decision. For a more thorough explanation of the initial values of the weights, check out Andrew Ng (th-cam.com/video/6by6Xas_Kho/w-d-xo.html). Good luck!
Thanks for this! clear and straight to the point. The simplest example I've seen so far. As regards biases, does this apply to the RBF Neural Networks as well? For instance, if the input layer has a bias unit, do we add the bias during forward propagation? Thanks
What is the point of saying "okay" every three seconds? Imagine people listening to your videos multiple times, it makes a lot of "okay, okay, okay, okay, okay". Which is just "noise" in the information that's actually blocking the material!
You may learn how to bring across your message in a more pleasant manner. We get an excellent lesson for free. And you are focusing on a minor issue. Think about it, if you can.
@@joachimguth6226 You are right, it could have been sugarcoated more. Unfortunately this feedback to benefit and improve the video/channel is answered with insult and disrespect. When a teacher starts insulting the audience with disrespect, it's a clear sign that he/she isn't the right character to be a teacher, but just uses information as an excuse to feel above others.
Very Good Very Nice
yes
such an clear explaination, this is truly a gift from the internet, thank you Bevan
Beautiful video. The patience, the calmness in your voice in these videos. Thank you so much
This is the best video on forward propagation in TH-cam. Thanks for posting such videos.
I 've never seen such a perfect explanation.I have faced with a problem while I was writing my kernel and searched everywhere to find a solution for understanding reason of error I ve got.Suddenly I saw the video and started to listen carefully .Thank you so much sir
I'm very happy to hear this :) Good luck to you
Part 1 and Part 2 are the best explanation of these subjects going. Thanks so much Bevan. You have a real talent for explaining a difficult subject in a way that’s as easy as possible to grasp. Brilliant videos
You're very welcome!
Most amazing explanation
Thanks a lot Bevan. Excellent insights with simplicity. Kudos!
Simply the best explanations i have found if i could give part one and two a hundred likes i would...Please do more AI content you have a gift for teaching.
the best explanation so far...
Thank you so much. This the first explanation of forward propagation in neural networks, that I actually understood.
Awesome to hear. Good luck!
Excellent presentation! I wish a intro in Transformer Architecture, which today replaces many NN (CNN,RNN,…).
Thank you so much for such a simple way to teach a medium student. It was too clear and obvious to understand and learn the NN.. Love from Pakistan😀
so perfect brother nice job
The best content I have found, thanks for such a detailed explanation!
Glad it was helpful!
I'm currently taking an AI course but I.must say your explanation is.more understandable. Thank you.
Dear Bevan,
You have presented both the forward and backward propagation concepts in an exceptional manner, especially the example that you have chosen for doing so. The majority of ANN tutorials stick to 'cat-dog' kind of categorical examples.
I have one query though. Do you recommend the use of a non-linear activation function in the output layer? And if yes, which non-linear activation function will you prefer for continuous problems. The thing to keep in mind while choosing this output layer activation function is that it should be able to provide an output which can exceed 1 or can be less than 0 or -1. Sigmoid or tanh activation function in the output layer will not allow us to have such 'greater than 1 or less than 0 / -1' outputs.
I hope to hear from you. Thanks!
Thank You. Very Informative
Awesome
Great work
Thankyou
My pleasure Gouri
This is a fantastic explanation, but you'll need some knowledge of machine learning from scratch, particularly linear and logistic regression. This is perfect and exactly what I needed to understand NN. Liked and subbed.
Is it possible for you to upload the slides for the series please?
Stochastic Gradient Decent. It is Stochastic since it Randomly shuffle samples in the training set.
May I ask how you can determine the bias to be added in z1 calculation?
Thanks for your explanation! How do you choose the weights? or its a random decision? Thank u in advance!
Hi Lucia. Yes it was a random decision. For a more thorough explanation of the initial values of the weights, check out Andrew Ng (th-cam.com/video/6by6Xas_Kho/w-d-xo.html). Good luck!
Why haven't we performed Activation function on Ypred ??
I was wondering why didn't you write the active fucnton in your output neuron? is there a specific reason?
Most likely because it is just a linear summation of the inputs. There was no need to pass it through a non linear activation function
Why do we only use 2 nodes?
It is just an example. There can be as many nodes/neurons as you wish. I just used two for this example
Thanks for this! clear and straight to the point. The simplest example I've seen so far.
As regards biases, does this apply to the RBF Neural Networks as well? For instance, if the input layer has a bias unit, do we add the bias during forward propagation? Thanks
What is the point of saying "okay" every three seconds?
Imagine people listening to your videos multiple times, it makes a lot of "okay, okay, okay, okay, okay". Which is just "noise" in the information that's actually blocking the material!
Then don't watch. Go elsewhere
@@bevansmithdatascience9580 Ignorance is bliss. Thanks, but no one is asking you what they should do next.
@@ChargedPulsar I'm telling you
You may learn how to bring across your message in a more pleasant manner. We get an excellent lesson for free. And you are focusing on a minor issue. Think about it, if you can.
@@joachimguth6226 You are right, it could have been sugarcoated more. Unfortunately this feedback to benefit and improve the video/channel is answered with insult and disrespect. When a teacher starts insulting the audience with disrespect, it's a clear sign that he/she isn't the right character to be a teacher, but just uses information as an excuse to feel above others.