Back propagation equation proof
$begingroup$
I am trying to prove this equation (from the backpropagation equations in AI).
$$frac{partial C}{partial b_j^l} = delta_j^l$$
C is the cost function: $C = frac{1}{2}||y - a^L||^2$
Where the output of layer l and neuron j is express like so $a^l_j=σ(∑_kw^l_{jk}a^{l−1}_k+b^l_j)$
I am suppose to use this assertion to do the demonstration: $delta_j^L = frac{partial C}{partial z_j^L}$
So far, here is what I have tried:
$$frac{partial C}{partial z_j^L} = sum_k frac{partial C}{partial z_j^L} frac{partial b_j^l}{partial b_j^L} $$ (I am using the chain rule to have a sum)
<=> $$frac{partial C}{partial z_j^L} = sum_k frac{partial C}{partial b_j^l} frac{partial b_j^l}{partial z_j^L} $$
So I guess I have to prove, that $frac{partial b_j^l}{partial z_j^L}$ equals 1.
But I don't have any ideas how to prove it.
Thanks for your help
N.B
I am following this course => http://neuralnetworksanddeeplearning.com/chap2.html where the first two equations of the BackPropagation equations are already proved, and the 2 others should be proved the same way (using the chain rule)
artificial-intelligence
$endgroup$
add a comment |
$begingroup$
I am trying to prove this equation (from the backpropagation equations in AI).
$$frac{partial C}{partial b_j^l} = delta_j^l$$
C is the cost function: $C = frac{1}{2}||y - a^L||^2$
Where the output of layer l and neuron j is express like so $a^l_j=σ(∑_kw^l_{jk}a^{l−1}_k+b^l_j)$
I am suppose to use this assertion to do the demonstration: $delta_j^L = frac{partial C}{partial z_j^L}$
So far, here is what I have tried:
$$frac{partial C}{partial z_j^L} = sum_k frac{partial C}{partial z_j^L} frac{partial b_j^l}{partial b_j^L} $$ (I am using the chain rule to have a sum)
<=> $$frac{partial C}{partial z_j^L} = sum_k frac{partial C}{partial b_j^l} frac{partial b_j^l}{partial z_j^L} $$
So I guess I have to prove, that $frac{partial b_j^l}{partial z_j^L}$ equals 1.
But I don't have any ideas how to prove it.
Thanks for your help
N.B
I am following this course => http://neuralnetworksanddeeplearning.com/chap2.html where the first two equations of the BackPropagation equations are already proved, and the 2 others should be proved the same way (using the chain rule)
artificial-intelligence
$endgroup$
add a comment |
$begingroup$
I am trying to prove this equation (from the backpropagation equations in AI).
$$frac{partial C}{partial b_j^l} = delta_j^l$$
C is the cost function: $C = frac{1}{2}||y - a^L||^2$
Where the output of layer l and neuron j is express like so $a^l_j=σ(∑_kw^l_{jk}a^{l−1}_k+b^l_j)$
I am suppose to use this assertion to do the demonstration: $delta_j^L = frac{partial C}{partial z_j^L}$
So far, here is what I have tried:
$$frac{partial C}{partial z_j^L} = sum_k frac{partial C}{partial z_j^L} frac{partial b_j^l}{partial b_j^L} $$ (I am using the chain rule to have a sum)
<=> $$frac{partial C}{partial z_j^L} = sum_k frac{partial C}{partial b_j^l} frac{partial b_j^l}{partial z_j^L} $$
So I guess I have to prove, that $frac{partial b_j^l}{partial z_j^L}$ equals 1.
But I don't have any ideas how to prove it.
Thanks for your help
N.B
I am following this course => http://neuralnetworksanddeeplearning.com/chap2.html where the first two equations of the BackPropagation equations are already proved, and the 2 others should be proved the same way (using the chain rule)
artificial-intelligence
$endgroup$
I am trying to prove this equation (from the backpropagation equations in AI).
$$frac{partial C}{partial b_j^l} = delta_j^l$$
C is the cost function: $C = frac{1}{2}||y - a^L||^2$
Where the output of layer l and neuron j is express like so $a^l_j=σ(∑_kw^l_{jk}a^{l−1}_k+b^l_j)$
I am suppose to use this assertion to do the demonstration: $delta_j^L = frac{partial C}{partial z_j^L}$
So far, here is what I have tried:
$$frac{partial C}{partial z_j^L} = sum_k frac{partial C}{partial z_j^L} frac{partial b_j^l}{partial b_j^L} $$ (I am using the chain rule to have a sum)
<=> $$frac{partial C}{partial z_j^L} = sum_k frac{partial C}{partial b_j^l} frac{partial b_j^l}{partial z_j^L} $$
So I guess I have to prove, that $frac{partial b_j^l}{partial z_j^L}$ equals 1.
But I don't have any ideas how to prove it.
Thanks for your help
N.B
I am following this course => http://neuralnetworksanddeeplearning.com/chap2.html where the first two equations of the BackPropagation equations are already proved, and the 2 others should be proved the same way (using the chain rule)
artificial-intelligence
artificial-intelligence
asked Nov 18 '18 at 13:41
UnepierreUnepierre
62
62
add a comment |
add a comment |
1 Answer
1
active
oldest
votes
$begingroup$
By definition, $$ frac{partial C}{partial z_j^ell} = delta_j^ell $$ $$ z^ell_k = sum_i w_{ki}^ell a^{ell -1}_j + b_k^{ell} $$
the latter of which means that
$$ frac{partial z^ell_k}{partial b_alpha^ell} = Delta^ell_{kalpha}=begin{cases}
1 & text{if } k=alpha \ 0 & text{otherwise} end{cases} $$
We want to show that
$$ frac{partial C}{partial b_j^ell} = delta_j^ell $$
Since $C$ depends on $b_beta$ through $z^beta$, we can use the chain rule:
begin{align}
frac{partial C}{partial b_j^ell}
&= sum_zeta frac{partial C}{partial z_zeta^ell} frac{partial z_zeta^ell}{partial b_j^ell} \
&= sum_zeta delta^ell_zeta Delta^ell_{zeta j} \
&= delta^ell_j
end{align}
where the Kronecker delta $Delta_{zeta j}^ell$ collapses out the sum except for the $j$th term.
$endgroup$
add a comment |
Your Answer
StackExchange.ready(function() {
var channelOptions = {
tags: "".split(" "),
id: "69"
};
initTagRenderer("".split(" "), "".split(" "), channelOptions);
StackExchange.using("externalEditor", function() {
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled) {
StackExchange.using("snippets", function() {
createEditor();
});
}
else {
createEditor();
}
});
function createEditor() {
StackExchange.prepareEditor({
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: true,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: 10,
bindNavPrevention: true,
postfix: "",
imageUploader: {
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
},
noCode: true, onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
});
}
});
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f3003546%2fback-propagation-equation-proof%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
1 Answer
1
active
oldest
votes
1 Answer
1
active
oldest
votes
active
oldest
votes
active
oldest
votes
$begingroup$
By definition, $$ frac{partial C}{partial z_j^ell} = delta_j^ell $$ $$ z^ell_k = sum_i w_{ki}^ell a^{ell -1}_j + b_k^{ell} $$
the latter of which means that
$$ frac{partial z^ell_k}{partial b_alpha^ell} = Delta^ell_{kalpha}=begin{cases}
1 & text{if } k=alpha \ 0 & text{otherwise} end{cases} $$
We want to show that
$$ frac{partial C}{partial b_j^ell} = delta_j^ell $$
Since $C$ depends on $b_beta$ through $z^beta$, we can use the chain rule:
begin{align}
frac{partial C}{partial b_j^ell}
&= sum_zeta frac{partial C}{partial z_zeta^ell} frac{partial z_zeta^ell}{partial b_j^ell} \
&= sum_zeta delta^ell_zeta Delta^ell_{zeta j} \
&= delta^ell_j
end{align}
where the Kronecker delta $Delta_{zeta j}^ell$ collapses out the sum except for the $j$th term.
$endgroup$
add a comment |
$begingroup$
By definition, $$ frac{partial C}{partial z_j^ell} = delta_j^ell $$ $$ z^ell_k = sum_i w_{ki}^ell a^{ell -1}_j + b_k^{ell} $$
the latter of which means that
$$ frac{partial z^ell_k}{partial b_alpha^ell} = Delta^ell_{kalpha}=begin{cases}
1 & text{if } k=alpha \ 0 & text{otherwise} end{cases} $$
We want to show that
$$ frac{partial C}{partial b_j^ell} = delta_j^ell $$
Since $C$ depends on $b_beta$ through $z^beta$, we can use the chain rule:
begin{align}
frac{partial C}{partial b_j^ell}
&= sum_zeta frac{partial C}{partial z_zeta^ell} frac{partial z_zeta^ell}{partial b_j^ell} \
&= sum_zeta delta^ell_zeta Delta^ell_{zeta j} \
&= delta^ell_j
end{align}
where the Kronecker delta $Delta_{zeta j}^ell$ collapses out the sum except for the $j$th term.
$endgroup$
add a comment |
$begingroup$
By definition, $$ frac{partial C}{partial z_j^ell} = delta_j^ell $$ $$ z^ell_k = sum_i w_{ki}^ell a^{ell -1}_j + b_k^{ell} $$
the latter of which means that
$$ frac{partial z^ell_k}{partial b_alpha^ell} = Delta^ell_{kalpha}=begin{cases}
1 & text{if } k=alpha \ 0 & text{otherwise} end{cases} $$
We want to show that
$$ frac{partial C}{partial b_j^ell} = delta_j^ell $$
Since $C$ depends on $b_beta$ through $z^beta$, we can use the chain rule:
begin{align}
frac{partial C}{partial b_j^ell}
&= sum_zeta frac{partial C}{partial z_zeta^ell} frac{partial z_zeta^ell}{partial b_j^ell} \
&= sum_zeta delta^ell_zeta Delta^ell_{zeta j} \
&= delta^ell_j
end{align}
where the Kronecker delta $Delta_{zeta j}^ell$ collapses out the sum except for the $j$th term.
$endgroup$
By definition, $$ frac{partial C}{partial z_j^ell} = delta_j^ell $$ $$ z^ell_k = sum_i w_{ki}^ell a^{ell -1}_j + b_k^{ell} $$
the latter of which means that
$$ frac{partial z^ell_k}{partial b_alpha^ell} = Delta^ell_{kalpha}=begin{cases}
1 & text{if } k=alpha \ 0 & text{otherwise} end{cases} $$
We want to show that
$$ frac{partial C}{partial b_j^ell} = delta_j^ell $$
Since $C$ depends on $b_beta$ through $z^beta$, we can use the chain rule:
begin{align}
frac{partial C}{partial b_j^ell}
&= sum_zeta frac{partial C}{partial z_zeta^ell} frac{partial z_zeta^ell}{partial b_j^ell} \
&= sum_zeta delta^ell_zeta Delta^ell_{zeta j} \
&= delta^ell_j
end{align}
where the Kronecker delta $Delta_{zeta j}^ell$ collapses out the sum except for the $j$th term.
answered Jan 13 at 22:52
user3658307user3658307
5,0833949
5,0833949
add a comment |
add a comment |
Thanks for contributing an answer to Mathematics Stack Exchange!
- Please be sure to answer the question. Provide details and share your research!
But avoid …
- Asking for help, clarification, or responding to other answers.
- Making statements based on opinion; back them up with references or personal experience.
Use MathJax to format equations. MathJax reference.
To learn more, see our tips on writing great answers.
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f3003546%2fback-propagation-equation-proof%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown