
After finding the errors and deviations in the data values, we must adjust the appropriate value for the weights and biases through the derivative of the deviation function related to them, which indicates the slope of the function in calculus.
If the value of the derivative is known to us, we reset the ratios of weights and biases by raising or decreasing those ratios, but this is not enough to calculate the derivative of the skew function directly in terms of weights and biases because their value does not exist in the equation that represents them, so it is necessary to use the law of the serial scale to reach the solution .

This mathematical equation may seem somewhat complicated, but it is the only way to lead us to the correct solution. For simplicity, we have shown the partial derivative of a single-layer Neural Network.
After we extract this result, we add the backpropagation task in the Python code for our case.

And in the following video tutorial, a detailed explanation of the law of the serial scale in backpropagation and the application of calculus is provided by 3Blue1Brown
Backpropagation

ุจุนุฏ ุงูุนุซูุฑ ุนูู ุงูุฃุฎุทุงุก ูุงูุงูุญุฑุงูุงุช ูู ููู ุงูุจูุงูุงุช ูุชูุฌุจ ุนูููุง ุถุจุท ุงูููู ุฉ ุงูู ูุงุณุจุฉ ููุฃูุฒุงู ูุงูุชุญูุฒุงุช ู ู ุฎูุงู ู ุดุชู ุฏุงูุฉ ุงูุงูุญุฑุงู ุงูู ุชุนููุฉ ุจูู ุง ูุงูุชู ุชุฏู ุนูู ู ูู ุงูุฏุงูุฉ ุจุญุณุงุจ ุงูุชูุงุถู ูุงูุชูุงู ู

ุฅุฐุง ูุงูุช ููู ุฉ ุงูู ุดุชู ู ุนููู ุฉ ูุฏููุง ููููู ุจุฅุนุงุฏุฉ ุชุนููู ูุณุจ ุงูุฃูุฒุงู ูุงูุชุญูุฒุงุช ุนู ุทุฑูู ุฑูุน ุชูู ุงููุณุจ ุฃู ุฎูุถูุง ุฅูุง ุฃู ูุฐุง ูุง ูููู ูุญุณุงุจ ู ุดุชู ุฏุงูุฉ ุงูุงูุญุฑุงู ู ุจุงุดุฑุฉู ุจุฏูุงูุฉ ุงูุฃูุฒุงู ูุงูุชุญูุฒุงุช ูุฃู ููู ุชูู ุง ุบูุฑ ู ูุฌูุฏุฉ ูู ุงูู ุนุงุฏูุฉ ุงูู ู ุซูุฉ ููุง ูุฐูู ูุงุจุฏ ู ู ุงูุงุณุชุนุงูุฉ ุจูุงููู ุงูู ููุงุณ ุงูุชุณูุณูู ูููุตูู ุฅูู ุงูุญู

ูุฏ ุชุจุฏู ูุฐู ุงูู ุนุงุฏูุฉ ุงูุญุณุงุจูุฉ ู ุนูุฏุฉ ููุนุงู ู ุง ููููุง ุงูุณุจูู ุงููุญูุฏ ูุฅูุตุงููุง ุฅูู ุงูุญู ุงูุณููู ูุนูู ุณุจูู ุงูุชุจุณูุท ูู ูุง ุฅุธูุงุฑ ุงูู ุดุชู ุงูุฌุฒุฆู
ู ุคููุฉ ู ู ุทุจูุฉ ูุงุญุฏุฉ Neural Network ูู
ูุจุนุฏ ุงุณุชุฎูุงุตูุง ููุฐู ุงููุชูุฌุฉ ูููู ุจุฅุถุงูุฉ
ูู ููุฏุงุช ุจุงูุซูู ูุญุงูุชูุง ูุฐู backpropagation ู ูู ุฉ

ููู ุงูููุฏูู ุงูุชุนููู ู ุงูุชุงูู ุดุฑุญ ู ูุตู ููุงููู
backpropagation ุงูู ููุงุณ ุงูุชุณูุณูู ูู
Blue1Brown3 ูุชุทุจูู ุงูุชูุงุถู ูุงูุชูุงู ู ู ูุฏู ู ู


You must be logged in to post a comment.