Methods and systems for text sequence style transfer by two encoder decoders

Patent metadata

Field	Value
Publication number	US-11501159-B2
Application number	US-201916365637-A
Country	US
Kind code	B2
Filing date	Mar 26, 2019
Priority date	Mar 26, 2019
Publication date	Nov 15, 2022
Grant date	Nov 15, 2022

Abstract

A method for text sequence style transfer by two encoder-decoders, including generating, by a first encoder-decoder network model, an output sequence based on a first input sequence and an input sequence style, wherein the output sequence is associated with a second sequence, generating, by a second-encoder decoder network model, a prediction of the first input sequence based on the first input sequence, the output sequence, and a first input sequence style associated with the first input sequence, generating, by a classifier, a prediction of the first input sequence style based on the prediction of the first input sequence, and updating the neural network model based on comparisons between the output sequence and the second sequence, between the prediction of the first input sequence and the first input sequence, and between the prediction of the first input sequence style and the first input sequence style.

First claim

What is claimed is:

1. A method for training a neural network model comprising:
generating, by a first encoder-decoder network model, an output sequence based on a first input sequence and an input sequence style, wherein the output sequence is associated with a second sequence;
generating, by a second-encoder decoder network model, a prediction of the first input sequence based on the first input sequence, the output sequence, and a first input sequence style associated with the first input sequence;
generating, by a classifier, a prediction of the first input sequence style based on the prediction of the first input sequence; and
updating the neural network model based on comparisons between the output sequence and the second sequence, between the prediction of the first input sequence and the first input sequence, and between the prediction of the first input sequence style and the first input sequence style.

2. The method of claim 1 , wherein the neural network model is a sequence to sequence neural network model.

3. The method of claim 1 , wherein the first encoder-decoder network model and the second encoder-decoder network model each includes at least two recursive neural network models.

4. The method of claim 3 , wherein each of the at least two recursive neural network models include one or more long short-term memory models.

5. The method of claim 1 , wherein updating the neural network model based on comparisons between the output sequence and the second sequence, between the prediction of the first input sequence and the first input sequence, and between the prediction of the first input sequence style and the first input sequence style comprises:
determining a first loss based on a comparison between the output sequence and the second sequence;
determining a second loss based on a comparison between the prediction of the first input sequence and the first input sequence;
determining a third loss based on a comparison between the prediction of the first input sequence style and the first input sequence style; and
determining a total loss based on the first, second, and third losses.

6. The method of claim 1 , further comprising updating parameters of the neural network model based on the total loss.

7. The method of claim 1 , wherein the first input sequence, the second sequence, and the first input sequence style are known prior to the training.

8. A method for text sequence style transfer by two encoder decoders of a trained neural network model comprising:
generating, by a trained first encoder-decoder network model, an output sequence based on a first input sequence and a first input sequence style, wherein the output sequence is associated with a second sequence; and
generating, by a trained second encoder-decoder network model, a target sequence based on the first input sequence, the output sequence, and a target sequence style associated with the target sequence.

9. The method of claim 8 , wherein the trained neural network model is a sequence to sequence neural network model.

10. The method of claim 8 , wherein the trained first encoder-decoder network model and the trained second encoder-encoder network model each includes at least two recursive neural network models.

11. The method of claim 10 , wherein each of the at least two recursive neural network models include one or more long short-term memory models.

12. A system for training a neural network model comprising:
a first encoder-decoder network model configured to generate an output sequence based on a first input sequence and an input sequence style, wherein the output sequence is associated with a second sequence;
a second encoder-decoder network model configured to generate a prediction of the first input sequence based on the first input sequence, the output sequence, and a first input sequence style associated with the first input sequence; and
a classifier configured to generate a prediction of the first input sequence style based on the prediction of the first input sequence; wherein the neural network model is updated based on comparisons between the output sequence and the second sequence, between the prediction of the first input sequence and the first input sequence, and between the prediction of the first input sequence style and the first input sequence style.

13. The system of claim 12 , wherein the neural network model is a sequence to sequence neural network model.

14. The system of claim 12 , wherein the first encoder-decoder network model and the second encoder-decoder network model each includes at least two recursive neural network models.

15. The system of claim 14 , wherein the at least two recursive neural network models include one or more long short-term memory models.

16. The system of claim 12 , wherein updating the neural network model based on comparisons between the output sequence and the second sequence, between the prediction of the first input sequence and the first input sequence, and between the prediction of the first input sequence style and the first input sequence style comprises:
determining a first loss based on a comparison between the output sequence and the second sequence;
determining a second loss based on a comparison between the prediction of the first input sequence and the first input sequence;
determining a third loss based on a comparison between the prediction of the first input sequence style and the first input sequence style; and
determining a total loss based on the first, second, and third losses.

17. The system of claim 12 , wherein updating the neural network model based on comparisons between the output sequence and the second sequence, between the prediction of the first input sequence and the first input sequence, and between the prediction of the first input sequence style and the first input sequence style further comprises: updating the parameters of the neural network model based on the total loss.

18. The system of claim 12 , wherein the first input sequence, the second sequence, and the first input sequence style are known prior to the training.

Assignees

• Alibaba Group Holding Ltd

Inventors

• Wu Shanchan
• Zhang Qiong
• Si Luo

Classifications

• G06N3/08 Primary

  Learning methods · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title

• G06N3/063

  using electronic means · CPC title

• G06N3/0454

  Physics · mapped topic