Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Next revision
Previous revision
summarization [2017/08/13 17:30]
127.0.0.1 external edit
summarization [2018/11/11 10:28]
admin
Line 45: Line 45:
 We address the problem of end-to-end visual storytelling. Given a photo album, our model first selects the most representative (summary) photos, and then composes a natural language story for the album. For this task, we make use of the Visual Storytelling dataset and a model composed of three hierarchically-attentive Recurrent Neural Nets (RNNs) to: encode the album photos, select representative (summary) photos, and compose the story. Automatic and human evaluations show our model achieves better performance on selection, generation, and retrieval than baselines. We address the problem of end-to-end visual storytelling. Given a photo album, our model first selects the most representative (summary) photos, and then composes a natural language story for the album. For this task, we make use of the Visual Storytelling dataset and a model composed of three hierarchically-attentive Recurrent Neural Nets (RNNs) to: encode the album photos, select representative (summary) photos, and compose the story. Automatic and human evaluations show our model achieves better performance on selection, generation, and retrieval than baselines.
  
 +https://​arxiv.org/​abs/​1801.10198 Generating Wikipedia by Summarizing Long Sequences
 +
 +https://​arxiv.org/​abs/​1804.05685v1 A Discourse-Aware Attention Model for Abstractive Summarization of Long Documents
 +
 +https://​arxiv.org/​abs/​1808.10792 Bottom-Up Abstractive Summarization
 +
 +https://​arxiv.org/​abs/​1810.05739 Unsupervised Neural Multi-document Abstractive Summarization
 +
 +https://​arxiv.org/​pdf/​1811.01824.pdf STRUCTURED NEURAL SUMMARIZATION
 +
 + Based
 +on the promising results of graph neural networks on highly structured data, we develop
 +a framework to extend existing sequence encoders with a graph component
 +that can reason about long-distance relationships in weakly structured data such as
 +text. In an extensive evaluation, we show that the resulting hybrid sequence-graph
 +models outperform both pure sequence models as well as pure graph models on a
 +range of summarization tasks.
 +
 +We presented a framework for extending sequence encoders with a graph component that can leverage
 +rich additional structure. In an evaluation on three different summarization tasks, we have shown
 +that this augmentation improves the performance of a range of different sequence models across all
 +tasks.