Loss function for a linear classifier












2












$begingroup$


I'm writing a loss function for a linear classifier. The function takes in several arrays and outputs the loss. s is a score array which contains 10 entries that are the scores for each class. y is the array that contains the correct class assignments. My loss function takes the max of 0 and the calculation between the correct score and all the other scores to see if there is a higher value in s than y[i].



import numpy as np
import math

def model_loss(W, b, x, y):
n = x.shape[0]
lossSize = 1/n
Li = 0.0
loss = 0.0
for i in range(n):
s = (np.dot(W.transpose(), x[i])) + b
for j in range (W.shape[1]):
if (j != y[i]):
Li += max(0.0, (s[j] - s[y[i]] + 1.0))
loss += Li
Li = 0.0
loss *= LossSize
return loss


This is a loss function for a linear classifier so W is an array of weights with the shape (16, 10) and x is the data I want to test. The shape of x is (10,000, 16) because it stores about 10,000 images. Each image contains a histogram of 16 entries. s is an array which contains 10 entries. The highest value of s is the class prediction. So if entry 3 on s has the highest value then 3 is the prediction class (these numbers could correspond to things like boat, car, building etc). y is the array that contains the correct class assignment. So y = [3, 8, 8 ...]. For each y value, it's value is treated as an index for s. So if s[y[i]] is the highest value then it means the linear classifier assigned the image to the right class. If it's not then it means the linear classifier assigned the wrong class to the image so the values from all the other entries in s will be compared with the y entry to see just how much loss was taken. Hence it's called a loss function.



This produces what I want but now I need to optimize it. My question is, is there a way to get rid of the two for loops since for loops can be rather slow and I feel like there is a more efficient way of doing this.










share|improve this question









New contributor




Brandon MacLeod is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.







$endgroup$












  • $begingroup$
    That is the entire function.
    $endgroup$
    – Brandon MacLeod
    39 mins ago










  • $begingroup$
    Ok I included the def.
    $endgroup$
    – Brandon MacLeod
    37 mins ago










  • $begingroup$
    Ok I updated my question.
    $endgroup$
    – Brandon MacLeod
    22 mins ago
















2












$begingroup$


I'm writing a loss function for a linear classifier. The function takes in several arrays and outputs the loss. s is a score array which contains 10 entries that are the scores for each class. y is the array that contains the correct class assignments. My loss function takes the max of 0 and the calculation between the correct score and all the other scores to see if there is a higher value in s than y[i].



import numpy as np
import math

def model_loss(W, b, x, y):
n = x.shape[0]
lossSize = 1/n
Li = 0.0
loss = 0.0
for i in range(n):
s = (np.dot(W.transpose(), x[i])) + b
for j in range (W.shape[1]):
if (j != y[i]):
Li += max(0.0, (s[j] - s[y[i]] + 1.0))
loss += Li
Li = 0.0
loss *= LossSize
return loss


This is a loss function for a linear classifier so W is an array of weights with the shape (16, 10) and x is the data I want to test. The shape of x is (10,000, 16) because it stores about 10,000 images. Each image contains a histogram of 16 entries. s is an array which contains 10 entries. The highest value of s is the class prediction. So if entry 3 on s has the highest value then 3 is the prediction class (these numbers could correspond to things like boat, car, building etc). y is the array that contains the correct class assignment. So y = [3, 8, 8 ...]. For each y value, it's value is treated as an index for s. So if s[y[i]] is the highest value then it means the linear classifier assigned the image to the right class. If it's not then it means the linear classifier assigned the wrong class to the image so the values from all the other entries in s will be compared with the y entry to see just how much loss was taken. Hence it's called a loss function.



This produces what I want but now I need to optimize it. My question is, is there a way to get rid of the two for loops since for loops can be rather slow and I feel like there is a more efficient way of doing this.










share|improve this question









New contributor




Brandon MacLeod is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.







$endgroup$












  • $begingroup$
    That is the entire function.
    $endgroup$
    – Brandon MacLeod
    39 mins ago










  • $begingroup$
    Ok I included the def.
    $endgroup$
    – Brandon MacLeod
    37 mins ago










  • $begingroup$
    Ok I updated my question.
    $endgroup$
    – Brandon MacLeod
    22 mins ago














2












2








2





$begingroup$


I'm writing a loss function for a linear classifier. The function takes in several arrays and outputs the loss. s is a score array which contains 10 entries that are the scores for each class. y is the array that contains the correct class assignments. My loss function takes the max of 0 and the calculation between the correct score and all the other scores to see if there is a higher value in s than y[i].



import numpy as np
import math

def model_loss(W, b, x, y):
n = x.shape[0]
lossSize = 1/n
Li = 0.0
loss = 0.0
for i in range(n):
s = (np.dot(W.transpose(), x[i])) + b
for j in range (W.shape[1]):
if (j != y[i]):
Li += max(0.0, (s[j] - s[y[i]] + 1.0))
loss += Li
Li = 0.0
loss *= LossSize
return loss


This is a loss function for a linear classifier so W is an array of weights with the shape (16, 10) and x is the data I want to test. The shape of x is (10,000, 16) because it stores about 10,000 images. Each image contains a histogram of 16 entries. s is an array which contains 10 entries. The highest value of s is the class prediction. So if entry 3 on s has the highest value then 3 is the prediction class (these numbers could correspond to things like boat, car, building etc). y is the array that contains the correct class assignment. So y = [3, 8, 8 ...]. For each y value, it's value is treated as an index for s. So if s[y[i]] is the highest value then it means the linear classifier assigned the image to the right class. If it's not then it means the linear classifier assigned the wrong class to the image so the values from all the other entries in s will be compared with the y entry to see just how much loss was taken. Hence it's called a loss function.



This produces what I want but now I need to optimize it. My question is, is there a way to get rid of the two for loops since for loops can be rather slow and I feel like there is a more efficient way of doing this.










share|improve this question









New contributor




Brandon MacLeod is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.







$endgroup$




I'm writing a loss function for a linear classifier. The function takes in several arrays and outputs the loss. s is a score array which contains 10 entries that are the scores for each class. y is the array that contains the correct class assignments. My loss function takes the max of 0 and the calculation between the correct score and all the other scores to see if there is a higher value in s than y[i].



import numpy as np
import math

def model_loss(W, b, x, y):
n = x.shape[0]
lossSize = 1/n
Li = 0.0
loss = 0.0
for i in range(n):
s = (np.dot(W.transpose(), x[i])) + b
for j in range (W.shape[1]):
if (j != y[i]):
Li += max(0.0, (s[j] - s[y[i]] + 1.0))
loss += Li
Li = 0.0
loss *= LossSize
return loss


This is a loss function for a linear classifier so W is an array of weights with the shape (16, 10) and x is the data I want to test. The shape of x is (10,000, 16) because it stores about 10,000 images. Each image contains a histogram of 16 entries. s is an array which contains 10 entries. The highest value of s is the class prediction. So if entry 3 on s has the highest value then 3 is the prediction class (these numbers could correspond to things like boat, car, building etc). y is the array that contains the correct class assignment. So y = [3, 8, 8 ...]. For each y value, it's value is treated as an index for s. So if s[y[i]] is the highest value then it means the linear classifier assigned the image to the right class. If it's not then it means the linear classifier assigned the wrong class to the image so the values from all the other entries in s will be compared with the y entry to see just how much loss was taken. Hence it's called a loss function.



This produces what I want but now I need to optimize it. My question is, is there a way to get rid of the two for loops since for loops can be rather slow and I feel like there is a more efficient way of doing this.







python performance numpy machine-learning clustering






share|improve this question









New contributor




Brandon MacLeod is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.











share|improve this question









New contributor




Brandon MacLeod is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.









share|improve this question




share|improve this question








edited 17 mins ago









200_success

129k15153415




129k15153415






New contributor




Brandon MacLeod is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.









asked 2 hours ago









Brandon MacLeodBrandon MacLeod

112




112




New contributor




Brandon MacLeod is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.





New contributor





Brandon MacLeod is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.






Brandon MacLeod is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.












  • $begingroup$
    That is the entire function.
    $endgroup$
    – Brandon MacLeod
    39 mins ago










  • $begingroup$
    Ok I included the def.
    $endgroup$
    – Brandon MacLeod
    37 mins ago










  • $begingroup$
    Ok I updated my question.
    $endgroup$
    – Brandon MacLeod
    22 mins ago


















  • $begingroup$
    That is the entire function.
    $endgroup$
    – Brandon MacLeod
    39 mins ago










  • $begingroup$
    Ok I included the def.
    $endgroup$
    – Brandon MacLeod
    37 mins ago










  • $begingroup$
    Ok I updated my question.
    $endgroup$
    – Brandon MacLeod
    22 mins ago
















$begingroup$
That is the entire function.
$endgroup$
– Brandon MacLeod
39 mins ago




$begingroup$
That is the entire function.
$endgroup$
– Brandon MacLeod
39 mins ago












$begingroup$
Ok I included the def.
$endgroup$
– Brandon MacLeod
37 mins ago




$begingroup$
Ok I included the def.
$endgroup$
– Brandon MacLeod
37 mins ago












$begingroup$
Ok I updated my question.
$endgroup$
– Brandon MacLeod
22 mins ago




$begingroup$
Ok I updated my question.
$endgroup$
– Brandon MacLeod
22 mins ago










0






active

oldest

votes











Your Answer





StackExchange.ifUsing("editor", function () {
return StackExchange.using("mathjaxEditing", function () {
StackExchange.MarkdownEditor.creationCallbacks.add(function (editor, postfix) {
StackExchange.mathjaxEditing.prepareWmdForMathJax(editor, postfix, [["\$", "\$"]]);
});
});
}, "mathjax-editing");

StackExchange.ifUsing("editor", function () {
StackExchange.using("externalEditor", function () {
StackExchange.using("snippets", function () {
StackExchange.snippets.init();
});
});
}, "code-snippets");

StackExchange.ready(function() {
var channelOptions = {
tags: "".split(" "),
id: "196"
};
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: false,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: null,
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
},
onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
});


}
});






Brandon MacLeod is a new contributor. Be nice, and check out our Code of Conduct.










draft saved

draft discarded


















StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fcodereview.stackexchange.com%2fquestions%2f212186%2floss-function-for-a-linear-classifier%23new-answer', 'question_page');
}
);

Post as a guest















Required, but never shown

























0






active

oldest

votes








0






active

oldest

votes









active

oldest

votes






active

oldest

votes








Brandon MacLeod is a new contributor. Be nice, and check out our Code of Conduct.










draft saved

draft discarded


















Brandon MacLeod is a new contributor. Be nice, and check out our Code of Conduct.













Brandon MacLeod is a new contributor. Be nice, and check out our Code of Conduct.












Brandon MacLeod is a new contributor. Be nice, and check out our Code of Conduct.
















Thanks for contributing an answer to Code Review 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.




draft saved


draft discarded














StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fcodereview.stackexchange.com%2fquestions%2f212186%2floss-function-for-a-linear-classifier%23new-answer', 'question_page');
}
);

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







Popular posts from this blog

How to make a Squid Proxy server?

Is this a new Fibonacci Identity?

19世紀