Large number of training steps results in poor performance in transfer learningGetting low accuracy with deep convolutional NN trained on Imagenet 2011Odd results for Image Recognition using AlexNet in Deep LearningPerforming Transfer Learning Caffe - Googlenet MIT Places ModelWhy does the validation performance have large fluctuations during deep CNN training?Keras validation accuracy much lower than training accuracy even with the same dataset for both training and validationValidation Accuracy stuck to 50% using keras VGG16 modelDifference between training and testing accuracy+ Tensorflow tutorialTensorflow model accuracyWhy model produces the best performance after the first epoch when my training loss decreases and the accuracy of the validation set increases?training small amount of data on the large capacity network
Party going through airport security at separate times?
Writing an ace/aro character?
Why does sleeping increase the execution time of an independent piece of code that is executed before/afterwards?
What does the multimeter dial do internally?
Why is a mixture of two normally distributed variables only bimodal if their means differ by at least two times the common standard deviation?
Why the Cauchy Distribution is so useful?
Is it better in terms of durability to remove card+battery or to connect to charger/computer via USB-C?
Would a Nikon FG 20 film SLR camera take pictures without batteries?
Why does the Antonov AN-225 not have any winglets?
Why do airports remove/realign runways?
How does Kaya's Ghostform interact with Elenda, the Dusk Rose?
Is it possible for a character at any level to cast all 44 Cantrips in one week without Magic Items?
What's it called when the bad guy gets eaten?
What factors could lead to bishops establishing monastic armies?
How to evaluate the performance of open source solver?
What are the effects of abstaining from eating a certain flavor?
Four ships at the ocean with the same distance
What was the profession 芸者 (female entertainer) called in Germany?
Why AI became applicable only after Nvidia's chips were available?
Shrinking padding of node with label options
US citizen traveling with Peruvian passport
Is it ok for parents to kiss and romance with each other while their 2- to 8-year-old child watches?
Write a function
Intern not wearing safety equipment; how could I have handled this differently?
Large number of training steps results in poor performance in transfer learning
Getting low accuracy with deep convolutional NN trained on Imagenet 2011Odd results for Image Recognition using AlexNet in Deep LearningPerforming Transfer Learning Caffe - Googlenet MIT Places ModelWhy does the validation performance have large fluctuations during deep CNN training?Keras validation accuracy much lower than training accuracy even with the same dataset for both training and validationValidation Accuracy stuck to 50% using keras VGG16 modelDifference between training and testing accuracy+ Tensorflow tutorialTensorflow model accuracyWhy model produces the best performance after the first epoch when my training loss decreases and the accuracy of the validation set increases?training small amount of data on the large capacity network
.everyoneloves__top-leaderboard:empty,.everyoneloves__mid-leaderboard:empty,.everyoneloves__bot-mid-leaderboard:empty margin-bottom:0;
I have a question. I have used transfer learning to retrain googlenet on my image classification problem. I have 80,000 images which belong to 14 categories. I set number of training steps equal to 200,000. I think the code provided by Tensorflow should have drop out implimented and it trains based on random shuffling of dataset and cross validation approach, and and I do not see any overfiting in training and classification curves, and I get high cross validation accuracy and high test accuracy but when I apply my model to new dataset then I get poor classification result. Anybodey know what is going on?Thanks!
deep-learning
asked Mar 25 at 22:38
hsn15051hsn15051
1201 silver badge15 bronze badges
add a comment |
I have a question. I have used transfer learning to retrain googlenet on my image classification problem. I have 80,000 images which belong to 14 categories. I set number of training steps equal to 200,000. I think the code provided by Tensorflow should have drop out implimented and it trains based on random shuffling of dataset and cross validation approach, and and I do not see any overfiting in training and classification curves, and I get high cross validation accuracy and high test accuracy but when I apply my model to new dataset then I get poor classification result. Anybodey know what is going on?Thanks!
deep-learning
asked Mar 25 at 22:38
hsn15051hsn15051
1201 silver badge15 bronze badges
- Are the two datasets inherently the same? Is there any possible huge difference you are simply overlooking? - Have you applied the same preprocessing steps to the second dataset? - Have you defined a proper performance metric?
– Mohammad Razzaghnoori
Mar 25 at 23:07
Both datasets are pretty similar. I have not applied any preprocessing step as mean subtraction is performed in the code provided by tensorflow. I have not defined metric but I simply used the retrain.py script by tensorflow for transfer learning and retraining inception-v2
– hsn15051
Mar 26 at 0:13
- "when I apply my model to new dataset then I get poor classification result" sounds like overfitting. - Have you tried to retrain the model using 60,000 of your 80,000 images, then test them on the other 20,000 ? If it works, then it might be the dataset difference issue mentioned by @MohammadRazzaghnoori ; if not, then you might need to investigate the retrain.py .
– user10253771
Mar 26 at 6:12
add a comment |
I have a question. I have used transfer learning to retrain googlenet on my image classification problem. I have 80,000 images which belong to 14 categories. I set number of training steps equal to 200,000. I think the code provided by Tensorflow should have drop out implimented and it trains based on random shuffling of dataset and cross validation approach, and and I do not see any overfiting in training and classification curves, and I get high cross validation accuracy and high test accuracy but when I apply my model to new dataset then I get poor classification result. Anybodey know what is going on?Thanks!
deep-learning
asked Mar 25 at 22:38
hsn15051hsn15051
1201 silver badge15 bronze badges
I have a question. I have used transfer learning to retrain googlenet on my image classification problem. I have 80,000 images which belong to 14 categories. I set number of training steps equal to 200,000. I think the code provided by Tensorflow should have drop out implimented and it trains based on random shuffling of dataset and cross validation approach, and and I do not see any overfiting in training and classification curves, and I get high cross validation accuracy and high test accuracy but when I apply my model to new dataset then I get poor classification result. Anybodey know what is going on?Thanks!
deep-learning
deep-learning
asked Mar 25 at 22:38
hsn15051hsn15051
1201 silver badge15 bronze badges
asked Mar 25 at 22:38
hsn15051hsn15051
1201 silver badge15 bronze badges
asked Mar 25 at 22:38
hsn15051hsn15051
1201 silver badge15 bronze badges
asked Mar 25 at 22:38
hsn15051hsn15051
1201 silver badge15 bronze badges
asked Mar 25 at 22:38
hsn15051hsn15051
1201 silver badge15 bronze badges
1201 silver badge15 bronze badges
- Are the two datasets inherently the same? Is there any possible huge difference you are simply overlooking? - Have you applied the same preprocessing steps to the second dataset? - Have you defined a proper performance metric?
– Mohammad Razzaghnoori
Mar 25 at 23:07
Both datasets are pretty similar. I have not applied any preprocessing step as mean subtraction is performed in the code provided by tensorflow. I have not defined metric but I simply used the retrain.py script by tensorflow for transfer learning and retraining inception-v2
– hsn15051
Mar 26 at 0:13
- "when I apply my model to new dataset then I get poor classification result" sounds like overfitting. - Have you tried to retrain the model using 60,000 of your 80,000 images, then test them on the other 20,000 ? If it works, then it might be the dataset difference issue mentioned by @MohammadRazzaghnoori ; if not, then you might need to investigate the retrain.py .
– user10253771
Mar 26 at 6:12
add a comment |
- Are the two datasets inherently the same? Is there any possible huge difference you are simply overlooking? - Have you applied the same preprocessing steps to the second dataset? - Have you defined a proper performance metric?
– Mohammad Razzaghnoori
Mar 25 at 23:07
Both datasets are pretty similar. I have not applied any preprocessing step as mean subtraction is performed in the code provided by tensorflow. I have not defined metric but I simply used the retrain.py script by tensorflow for transfer learning and retraining inception-v2
– hsn15051
Mar 26 at 0:13
- "when I apply my model to new dataset then I get poor classification result" sounds like overfitting. - Have you tried to retrain the model using 60,000 of your 80,000 images, then test them on the other 20,000 ? If it works, then it might be the dataset difference issue mentioned by @MohammadRazzaghnoori ; if not, then you might need to investigate the retrain.py .
– user10253771
Mar 26 at 6:12
- Are the two datasets inherently the same? Is there any possible huge difference you are simply overlooking? - Have you applied the same preprocessing steps to the second dataset? - Have you defined a proper performance metric?
– Mohammad Razzaghnoori
Mar 25 at 23:07
- Are the two datasets inherently the same? Is there any possible huge difference you are simply overlooking? - Have you applied the same preprocessing steps to the second dataset? - Have you defined a proper performance metric?
– Mohammad Razzaghnoori
Mar 25 at 23:07
Both datasets are pretty similar. I have not applied any preprocessing step as mean subtraction is performed in the code provided by tensorflow. I have not defined metric but I simply used the retrain.py script by tensorflow for transfer learning and retraining inception-v2
– hsn15051
Mar 26 at 0:13
Both datasets are pretty similar. I have not applied any preprocessing step as mean subtraction is performed in the code provided by tensorflow. I have not defined metric but I simply used the retrain.py script by tensorflow for transfer learning and retraining inception-v2
– hsn15051
Mar 26 at 0:13
- "when I apply my model to new dataset then I get poor classification result" sounds like overfitting. - Have you tried to retrain the model using 60,000 of your 80,000 images, then test them on the other 20,000 ? If it works, then it might be the dataset difference issue mentioned by @MohammadRazzaghnoori ; if not, then you might need to investigate the retrain.py .
– user10253771
Mar 26 at 6:12
- "when I apply my model to new dataset then I get poor classification result" sounds like overfitting. - Have you tried to retrain the model using 60,000 of your 80,000 images, then test them on the other 20,000 ? If it works, then it might be the dataset difference issue mentioned by @MohammadRazzaghnoori ; if not, then you might need to investigate the retrain.py .
– user10253771
Mar 26 at 6:12
add a comment |
0
active
oldest
votes
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%2f55347399%2flarge-number-of-training-steps-results-in-poor-performance-in-transfer-learning%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
Is this question similar to what you get asked at work? Learn more about asking and sharing private information with your coworkers using Stack Overflow for Teams.
Is this question similar to what you get asked at work? Learn more about asking and sharing private information with your coworkers using Stack Overflow for Teams.
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%2f55347399%2flarge-number-of-training-steps-results-in-poor-performance-in-transfer-learning%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
- Are the two datasets inherently the same? Is there any possible huge difference you are simply overlooking? - Have you applied the same preprocessing steps to the second dataset? - Have you defined a proper performance metric?
– Mohammad Razzaghnoori
Mar 25 at 23:07
Both datasets are pretty similar. I have not applied any preprocessing step as mean subtraction is performed in the code provided by tensorflow. I have not defined metric but I simply used the retrain.py script by tensorflow for transfer learning and retraining inception-v2
– hsn15051
Mar 26 at 0:13
- "when I apply my model to new dataset then I get poor classification result" sounds like overfitting. - Have you tried to retrain the model using 60,000 of your 80,000 images, then test them on the other 20,000 ? If it works, then it might be the dataset difference issue mentioned by @MohammadRazzaghnoori ; if not, then you might need to investigate the retrain.py .
– user10253771
Mar 26 at 6:12