Denoising autoencoder image captioning
In an approach to augmenting a caption dataset by leveraging a denoising autoencoder to sample and generate additional captions from the ground truth captions, one or more computer processors generate a plurality of new captions utilizing an autoencoder fed with one or more noisy captions, wherein the autoencoder is trained with a dataset comprising a plurality of ground truth captions. The one or more computer processors calculate an importance weight for each new caption in the plurality of generated new captions as compared to a plurality of associated ground truth captions based on a consensus metric. The one or more computer processors train a caption model with the generated plurality of new captions and associated calculated weights.
1. A computer-implemented method comprising:
responsive to a subsequent image captioning request, generating, by one or more computer processors, a plurality of new captions utilizing an autoencoder fed with one or more noisy captions, wherein the autoencoder is trained with a dataset comprising a plurality of ground truth captions, by:
reconstructing, by one or more computer processors, the one or more noisy captions into the plurality of new captions each comprising features and correlations contained in the one or more noisy captions and an original caption;
calculating, by one or more computer processors, an importance weight for each new caption in the plurality of generated new captions as compared to a plurality of associated ground truth captions based on a consensus metric; and
training, by one or more computer processors, a caption model with the generated plurality of new captions and associated calculated weights.
2. The method of claim 1 , further comprising:
generating, by one or more computer processors, a subsequent caption for a subsequent image utilizing the trained caption model.
3. The method of claim 1 , further comprising:
adding, by one or more computer processors, stochastic depth as noise to each ground truth caption in the plurality of ground truth captions within one or more hidden layers contained in the trained autoencoder.
4. The method of claim 1 , further comprising:
applying, by one or more computer processors, dropout regularization to each ground truth caption in the plurality of ground truth captions.
5. The method of claim 1 , wherein the consensus metric utilizes consensus-based image description evaluation.
6. The method of claim 1 , wherein the consensus metric is an average cosine similarity between one or more generated captions and one or more associated ground truth captions.
7. The method of claim 1 , wherein the autoencoder is a denoising autoencoder.
8. The method of claim 1 , wherein the trained caption model is a recurrent neural network.
9. A computer program product comprising:
one or more computer readable storage media and program instructions stored on the one or more computer readable storage media, the stored program instructions comprising:
responsive to a subsequent image captioning request, program instructions to generate a plurality of new captions utilizing an autoencoder fed with one or more noisy captions, wherein the autoencoder is trained with a dataset comprising a plurality of ground truth captions, by:
program instructions to reconstruct the one or more noisy captions into the plurality of new captions each comprising features and correlations contained in the one or more noisy captions and an original caption;
program instructions to calculate an importance weight for each new caption in the plurality of generated new captions as compared to a plurality of associated ground truth captions based on a consensus metric; and
program instructions to train a caption model with the generated plurality of new captions and associated calculated weights.
10. The computer program product of claim 9 , wherein the program instructions, stored on the one or more computer readable storage media, further comprise:
program instructions to generate a subsequent caption for a subsequent image utilizing the trained caption model.
11. The computer program product of claim 9 , wherein the program instructions, stored on the one or more computer readable storage media, further comprise:
program instructions to add stochastic depth as noise to each ground truth caption in the plurality of ground truth captions within one or more hidden layers contained in the trained autoencoder.
12. The computer program product of claim 9 , wherein the program instructions, stored on the one or more computer readable storage media, further comprise:
program instructions to apply dropout regularization to each ground truth caption in the plurality of ground truth captions.
13. The computer program product of claim 9 , wherein the consensus metric utilizes consensus-based image description evaluation.
14. The computer program product of claim 9 , wherein the consensus metric is an average cosine similarity between one or more generated captions and one or more associated ground truth captions.
15. A computer system comprising:
one or more computer processors;
one or more computer readable storage media; and
program instructions stored on the computer readable storage media for execution by at least one of the one or more processors, the stored program instructions comprising:
program instructions to generate a plurality of new captions utilizing an autoencoder fed with one or more noisy captions, wherein the autoencoder is trained with a dataset comprising a plurality of ground truth captions, by:
program instructions to reconstruct the one or more noisy captions into the plurality of new captions each comprising features and correlations contained in the one or more noisy captions and an original caption;
program instructions to calculate an importance weight for each new caption in the plurality of generated new captions as compared to a plurality of associated ground truth captions based on a consensus metric; and
program instructions to train a caption model with the generated plurality of new captions and associated calculated weights.
16. The computer system of claim 15 , wherein the program instructions, stored on the one or more computer readable storage media, further comprise:
program instructions to generate a subsequent caption for a subsequent image utilizing the trained caption model.
17. The computer system of claim 15 , wherein the program instructions, stored on the one or more computer readable storage media, further comprise:
program instructions to add stochastic depth as noise to each ground truth caption in the plurality of ground truth captions within one or more hidden layers contained in the trained autoencoder.
18. The computer system of claim 15 , wherein the program instructions, stored on the one or more computer readable storage media, further comprise:
program instructions to apply dropout regularization to each ground truth caption in the plurality of ground truth captions.
19. The computer system of claim 15 , wherein the consensus metric utilizes consensus-based image description evaluation.
20. The computer system of claim 15 , wherein the consensus metric is an average cosine similarity between one or more generated captions and one or more associated ground truth captions.