CNTK Loss and Error Metric function for multi label classification
.everyoneloves__top-leaderboard:empty,.everyoneloves__mid-leaderboard:empty,.everyoneloves__bot-mid-leaderboard:empty{ height:90px;width:728px;box-sizing:border-box;
}
Other than squared_error what other loss function / error function would I be able to use ?
I looked through https://cntk.ai/pythondocs/cntk.losses.html
and wasn't able to find anything that helps.
i found documentation for brain script but not in python
any help would be amazing :)
cntk
asked Nov 29 '18 at 6:19
zero corezero core
84
add a comment |
Other than squared_error what other loss function / error function would I be able to use ?
I looked through https://cntk.ai/pythondocs/cntk.losses.html
and wasn't able to find anything that helps.
i found documentation for brain script but not in python
any help would be amazing :)
cntk
asked Nov 29 '18 at 6:19
zero corezero core
84
add a comment |
Other than squared_error what other loss function / error function would I be able to use ?
I looked through https://cntk.ai/pythondocs/cntk.losses.html
and wasn't able to find anything that helps.
i found documentation for brain script but not in python
any help would be amazing :)
cntk
asked Nov 29 '18 at 6:19
zero corezero core
84
Other than squared_error what other loss function / error function would I be able to use ?
I looked through https://cntk.ai/pythondocs/cntk.losses.html
and wasn't able to find anything that helps.
i found documentation for brain script but not in python
any help would be amazing :)
cntk
cntk
asked Nov 29 '18 at 6:19
zero corezero core
84
asked Nov 29 '18 at 6:19
zero corezero core
84
asked Nov 29 '18 at 6:19
zero corezero core
84
asked Nov 29 '18 at 6:19
zero corezero core
84
asked Nov 29 '18 at 6:19
zero corezero core
84
84
add a comment |
add a comment |
2 Answers
2
active
oldest
votes
The best source of documentation (IMHO) is python documentation. If you need to write your own loss function I found this post very helpful. Try using sigmoid function at the output layer and binary cross entropy loss or cosine loss.
target = cntk.input_variable(input_dim)
loss = cntk.binary_cross_entropy(z, target)
This way your nodes will output probabilities independent of each other like [0.73, 0.02, 0.05, 0.26, 0.68].
answered Dec 5 '18 at 12:26
papadoble151papadoble151
448414
add a comment |
For multi-class classification, we typically use cross_entropy_with_softmax.
You are trying to attribute 2 or more class to every sample, then there's no native implementation in cntk
answered Nov 29 '18 at 10:16
snowflakesnowflake
263210
I don't cntk has a built in category_cross_entropy_with_softmax ? .. i couldn't find it in the documentation or in the code
– zero core
Nov 29 '18 at 15:30
its in the python API, not too sure about brainscript. I updated the answer with the link
– snowflake
Nov 29 '18 at 15:32
Oh yes but correct me if i am worng but that one is mainly for one hot encoded target ... from their example I would ideally want - C.cross_entropy_with_softmax([[1., 1., 50., 50.]], [[0., 0., 1., 1.]]).eval() ...... to result in 0. ..... but it would say 50 ( array([[-49.30685425]], dtype=float32) ).
– zero core
Nov 29 '18 at 19:03
For classification problems, you must one hot encode.
– snowflake
Nov 30 '18 at 18:31
i don't think you understand what i am trying to say
– zero core
Dec 1 '18 at 23:47
|
show 2 more comments
Your Answer
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: "1"
};
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
},
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%2fstackoverflow.com%2fquestions%2f53532947%2fcntk-loss-and-error-metric-function-for-multi-label-classification%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
2 Answers
2
active
oldest
votes
2 Answers
2
active
oldest
votes
active
oldest
votes
active
oldest
votes
The best source of documentation (IMHO) is python documentation. If you need to write your own loss function I found this post very helpful. Try using sigmoid function at the output layer and binary cross entropy loss or cosine loss.
target = cntk.input_variable(input_dim)
loss = cntk.binary_cross_entropy(z, target)
This way your nodes will output probabilities independent of each other like [0.73, 0.02, 0.05, 0.26, 0.68].
answered Dec 5 '18 at 12:26
papadoble151papadoble151
448414
add a comment |
The best source of documentation (IMHO) is python documentation. If you need to write your own loss function I found this post very helpful. Try using sigmoid function at the output layer and binary cross entropy loss or cosine loss.
target = cntk.input_variable(input_dim)
loss = cntk.binary_cross_entropy(z, target)
This way your nodes will output probabilities independent of each other like [0.73, 0.02, 0.05, 0.26, 0.68].
answered Dec 5 '18 at 12:26
papadoble151papadoble151
448414
add a comment |
The best source of documentation (IMHO) is python documentation. If you need to write your own loss function I found this post very helpful. Try using sigmoid function at the output layer and binary cross entropy loss or cosine loss.
target = cntk.input_variable(input_dim)
loss = cntk.binary_cross_entropy(z, target)
This way your nodes will output probabilities independent of each other like [0.73, 0.02, 0.05, 0.26, 0.68].
answered Dec 5 '18 at 12:26
papadoble151papadoble151
448414
The best source of documentation (IMHO) is python documentation. If you need to write your own loss function I found this post very helpful. Try using sigmoid function at the output layer and binary cross entropy loss or cosine loss.
target = cntk.input_variable(input_dim)
loss = cntk.binary_cross_entropy(z, target)
This way your nodes will output probabilities independent of each other like [0.73, 0.02, 0.05, 0.26, 0.68].
answered Dec 5 '18 at 12:26
papadoble151papadoble151
448414
edited Dec 5 '18 at 12:59
answered Dec 5 '18 at 12:26
papadoble151papadoble151
448414
answered Dec 5 '18 at 12:26
papadoble151papadoble151
448414
answered Dec 5 '18 at 12:26
papadoble151papadoble151
448414
448414
add a comment |
add a comment |
For multi-class classification, we typically use cross_entropy_with_softmax.
You are trying to attribute 2 or more class to every sample, then there's no native implementation in cntk
answered Nov 29 '18 at 10:16
snowflakesnowflake
263210
I don't cntk has a built in category_cross_entropy_with_softmax ? .. i couldn't find it in the documentation or in the code
– zero core
Nov 29 '18 at 15:30
its in the python API, not too sure about brainscript. I updated the answer with the link
– snowflake
Nov 29 '18 at 15:32
Oh yes but correct me if i am worng but that one is mainly for one hot encoded target ... from their example I would ideally want - C.cross_entropy_with_softmax([[1., 1., 50., 50.]], [[0., 0., 1., 1.]]).eval() ...... to result in 0. ..... but it would say 50 ( array([[-49.30685425]], dtype=float32) ).
– zero core
Nov 29 '18 at 19:03
For classification problems, you must one hot encode.
– snowflake
Nov 30 '18 at 18:31
i don't think you understand what i am trying to say
– zero core
Dec 1 '18 at 23:47
|
show 2 more comments
For multi-class classification, we typically use cross_entropy_with_softmax.
You are trying to attribute 2 or more class to every sample, then there's no native implementation in cntk
answered Nov 29 '18 at 10:16
snowflakesnowflake
263210
I don't cntk has a built in category_cross_entropy_with_softmax ? .. i couldn't find it in the documentation or in the code
– zero core
Nov 29 '18 at 15:30
its in the python API, not too sure about brainscript. I updated the answer with the link
– snowflake
Nov 29 '18 at 15:32
Oh yes but correct me if i am worng but that one is mainly for one hot encoded target ... from their example I would ideally want - C.cross_entropy_with_softmax([[1., 1., 50., 50.]], [[0., 0., 1., 1.]]).eval() ...... to result in 0. ..... but it would say 50 ( array([[-49.30685425]], dtype=float32) ).
– zero core
Nov 29 '18 at 19:03
For classification problems, you must one hot encode.
– snowflake
Nov 30 '18 at 18:31
i don't think you understand what i am trying to say
– zero core
Dec 1 '18 at 23:47
|
show 2 more comments
For multi-class classification, we typically use cross_entropy_with_softmax.
You are trying to attribute 2 or more class to every sample, then there's no native implementation in cntk
answered Nov 29 '18 at 10:16
snowflakesnowflake
263210
For multi-class classification, we typically use cross_entropy_with_softmax.
You are trying to attribute 2 or more class to every sample, then there's no native implementation in cntk
answered Nov 29 '18 at 10:16
snowflakesnowflake
263210
edited Dec 3 '18 at 9:06
answered Nov 29 '18 at 10:16
snowflakesnowflake
263210
answered Nov 29 '18 at 10:16
snowflakesnowflake
263210
answered Nov 29 '18 at 10:16
snowflakesnowflake
263210
263210
I don't cntk has a built in category_cross_entropy_with_softmax ? .. i couldn't find it in the documentation or in the code
– zero core
Nov 29 '18 at 15:30
its in the python API, not too sure about brainscript. I updated the answer with the link
– snowflake
Nov 29 '18 at 15:32
Oh yes but correct me if i am worng but that one is mainly for one hot encoded target ... from their example I would ideally want - C.cross_entropy_with_softmax([[1., 1., 50., 50.]], [[0., 0., 1., 1.]]).eval() ...... to result in 0. ..... but it would say 50 ( array([[-49.30685425]], dtype=float32) ).
– zero core
Nov 29 '18 at 19:03
For classification problems, you must one hot encode.
– snowflake
Nov 30 '18 at 18:31
i don't think you understand what i am trying to say
– zero core
Dec 1 '18 at 23:47
|
show 2 more comments
I don't cntk has a built in category_cross_entropy_with_softmax ? .. i couldn't find it in the documentation or in the code
– zero core
Nov 29 '18 at 15:30
its in the python API, not too sure about brainscript. I updated the answer with the link
– snowflake
Nov 29 '18 at 15:32
Oh yes but correct me if i am worng but that one is mainly for one hot encoded target ... from their example I would ideally want - C.cross_entropy_with_softmax([[1., 1., 50., 50.]], [[0., 0., 1., 1.]]).eval() ...... to result in 0. ..... but it would say 50 ( array([[-49.30685425]], dtype=float32) ).
– zero core
Nov 29 '18 at 19:03
For classification problems, you must one hot encode.
– snowflake
Nov 30 '18 at 18:31
i don't think you understand what i am trying to say
– zero core
Dec 1 '18 at 23:47
I don't cntk has a built in category_cross_entropy_with_softmax ? .. i couldn't find it in the documentation or in the code
– zero core
Nov 29 '18 at 15:30
I don't cntk has a built in category_cross_entropy_with_softmax ? .. i couldn't find it in the documentation or in the code
– zero core
Nov 29 '18 at 15:30
its in the python API, not too sure about brainscript. I updated the answer with the link
– snowflake
Nov 29 '18 at 15:32
its in the python API, not too sure about brainscript. I updated the answer with the link
– snowflake
Nov 29 '18 at 15:32
Oh yes but correct me if i am worng but that one is mainly for one hot encoded target ... from their example I would ideally want - C.cross_entropy_with_softmax([[1., 1., 50., 50.]], [[0., 0., 1., 1.]]).eval() ...... to result in 0. ..... but it would say 50 ( array([[-49.30685425]], dtype=float32) ).
– zero core
Nov 29 '18 at 19:03
Oh yes but correct me if i am worng but that one is mainly for one hot encoded target ... from their example I would ideally want - C.cross_entropy_with_softmax([[1., 1., 50., 50.]], [[0., 0., 1., 1.]]).eval() ...... to result in 0. ..... but it would say 50 ( array([[-49.30685425]], dtype=float32) ).
– zero core
Nov 29 '18 at 19:03
For classification problems, you must one hot encode.
– snowflake
Nov 30 '18 at 18:31
For classification problems, you must one hot encode.
– snowflake
Nov 30 '18 at 18:31
i don't think you understand what i am trying to say
– zero core
Dec 1 '18 at 23:47
i don't think you understand what i am trying to say
– zero core
Dec 1 '18 at 23:47
|
show 2 more comments
Thanks for contributing an answer to Stack Overflow!
- 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.
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%2fstackoverflow.com%2fquestions%2f53532947%2fcntk-loss-and-error-metric-function-for-multi-label-classification%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
