a neural representation of sketch drawings: Difference between revisions

From statwiki
Jump to navigation Jump to search
No edit summary
No edit summary
Line 8: Line 8:
The authors offer two motivations for their work:
The authors offer two motivations for their work:
# To translate between languages for which large parallel corpora does not exist
# To translate between languages for which large parallel corpora does not exist
=== Note: lalala ===


== Methodology ==


In linguistics, a corpus (plural corpora) or text corpus and structured set of texts (nowadays usually electronically stored and processed). They are used to do statistical analysis and hypothesis testing, checking occurrences or validating linguistic rules within a specific language territory. A corpus may contain texts in a single language (monolingual corpus) or text data in multiple language (multilingual corpus).
=== Dataset ===


== Methodology ==
=== Sketch-RNN ===
The unsupervised translation scheme has the following outline:
* The word-vector embeddings of the source and target languages are aligned in an unsupervised manner.
* Sentences from the source and target language are mapped to a common latent vector space by an encoder, and then mapped to probability distributions over


==Encoder ==
=== Unconditional Generation ===
The encoder <math display="inline">E </math>  reads a sequence of word vectors <math display="inline">(z_1,\ldots, z_m) \in \mathcal{Z}'</math> and outputs a


==Decoder==
=== Training ===


The decoder is a mono-directional LSTM that accepts a sequence of hidden states <math display="inline">h=(h_1,\ldots, h_m) \in H'</math> from the latent space and a
== Experiments ==


[[File:paper4_fig2.png|700px|]]
=== Conditional Reconstruction ===


==Overview of objective ==
=== Latent Space Interpolation ===
The objective function is the sum of:
=== Sketch Drawing Analogies ===
# The de-noising auto-encoder loss,
=== Predicting Different Endings of Incomplete Sketches ===


I shall describe these in the following sections.
== Applications and Future Work ==


==De-noising Auto-encoder Loss ==  
== Conclusion ==
A de-noising auto-encoder is a function optimized to map a corrupted sample from some dataset to the original un-corrupted sample. De-noising auto-encoders were 


==Translation Loss==
== References ==  
To compute the translation loss, we sample a sentence from one of the languages, translate it with the encoder and decoder of the previous epoch, and then corrupt its
#Bahdanau, Dzmitry, Kyunghyun Cho, and Yoshua Bengio. "Neural machine translation by jointly learning to align and translate." arXiv preprint arXiv:1409.0473 (2014).


==Adversarial Loss ==
The intuition underlying the latent space is that it should encode the meaning of a sentence in a language-independent way.  Accordingly, the authors introduce an


==Objective Function==


Combining the above-described terms, we can write the overall objective function. Let <math display="inline">Q_S</math>  denote the monolingual dataset for the  
== fonts and examples ==
The unsupervised translation scheme has the following outline:
* The word-vector embeddings of the source and target languages are aligned in an unsupervised manner.
* Sentences from the source and target language are mapped to a common latent vector space by an encoder, and then mapped to probability distributions over


==Validation==
The objective function is the sum of:
The authors' aim is for their method to be completely unsupervised, so they do not use parallel corpora even for the selection of hyper-parameters.  Instead, they validate
# The de-noising auto-encoder loss,
==Experimental Procedure and Results==
I shall describe these in the following sections.


The authors test their method on four data sets. The first is from the English-French translation task of the Workshop on Machine Translation 2014 (WMT14).  This data set
[[File:paper4_fig2.png|700px|]]


==Result Figures==
[[File:MC_Translation Results.png]]
[[File:MC_Translation Results.png]]
==Commentary==
This paper's results are impressive: that it is even possible to translate between languages without parallel data suggests that languages are more similar than we might


[[File:MC_Alignment_Results.png|frame|none|alt=Alt text|From Conneau et al. (2017).  The final row shows the performance of alignment method used in the present paper.  Note the degradation in performance for more distant languages.]]
[[File:MC_Alignment_Results.png|frame|none|alt=Alt text|From Conneau et al. (2017).  The final row shows the performance of alignment method used in the present paper.  Note the degradation in performance for more distant languages.]]


[[File:MC_Translation_Ablation.png|frame|none|alt=Alt text|From the present paper. Results of an ablation study.  Of note are the first, third, and forth rows, which demonstrate that while the translation component of the loss is relatively unimportant, the word vector alignment scheme and de-noising auto-encoder matter a great deal.]]
[[File:MC_Translation_Ablation.png|frame|none|alt=Alt text|From the present paper. Results of an ablation study.  Of note are the first, third, and forth rows, which demonstrate that while the translation component of the loss is relatively unimportant, the word vector alignment scheme and de-noising auto-encoder matter a great deal.]]
==Future Work==
The  principal of performing unsupervised translation by starting with a rough but reasonable guess, and then improving it using knowledge of the structure of target
==References==
#Bahdanau, Dzmitry, Kyunghyun Cho, and Yoshua Bengio. "Neural machine translation by jointly learning to align and translate." arXiv preprint arXiv:1409.0473 (2014).

Revision as of 17:16, 15 November 2018

Introduction

lalala

  1. To provide a strong lower bound that any semi-supervised machine translation system is supposed to yield


Related Work

The authors offer two motivations for their work:

  1. To translate between languages for which large parallel corpora does not exist

Methodology

Dataset

Sketch-RNN

Unconditional Generation

Training

Experiments

Conditional Reconstruction

Latent Space Interpolation

Sketch Drawing Analogies

Predicting Different Endings of Incomplete Sketches

Applications and Future Work

Conclusion

References

  1. Bahdanau, Dzmitry, Kyunghyun Cho, and Yoshua Bengio. "Neural machine translation by jointly learning to align and translate." arXiv preprint arXiv:1409.0473 (2014).


fonts and examples

The unsupervised translation scheme has the following outline:

  • The word-vector embeddings of the source and target languages are aligned in an unsupervised manner.
  • Sentences from the source and target language are mapped to a common latent vector space by an encoder, and then mapped to probability distributions over

The objective function is the sum of:

  1. The de-noising auto-encoder loss,

I shall describe these in the following sections.

Alt text
From Conneau et al. (2017). The final row shows the performance of alignment method used in the present paper. Note the degradation in performance for more distant languages.
Alt text
From the present paper. Results of an ablation study. Of note are the first, third, and forth rows, which demonstrate that while the translation component of the loss is relatively unimportant, the word vector alignment scheme and de-noising auto-encoder matter a great deal.
Retrieved from "http://wiki.math.uwaterloo.ca/statwiki/index.php?title=a_neural_representation_of_sketch_drawings&oldid=39400"