NECLLGFeb 27, 2017

Improved Variational Autoencoders for Text Modeling using Dilated Convolutions

arXiv:1702.08139v2401 citations
Originality Incremental advance
AI Analysis

This addresses a key bottleneck in text generative modeling for researchers, providing the first positive experimental result for VAE in this domain, though it is incremental as it modifies the decoder architecture.

The paper tackled the problem of variational autoencoders (VAE) underperforming compared to LSTM language models in text generative modeling by proposing a dilated CNN decoder to control context and improve conditioning. It demonstrated perplexity gains on two datasets, showing VAE can outperform LSTM models, and achieved improvements in semi-supervised and unsupervised labeling tasks.

Recent work on generative modeling of text has found that variational auto-encoders (VAE) incorporating LSTM decoders perform worse than simpler LSTM language models (Bowman et al., 2015). This negative result is so far poorly understood, but has been attributed to the propensity of LSTM decoders to ignore conditioning information from the encoder. In this paper, we experiment with a new type of decoder for VAE: a dilated CNN. By changing the decoder's dilation architecture, we control the effective context from previously generated words. In experiments, we find that there is a trade off between the contextual capacity of the decoder and the amount of encoding information used. We show that with the right decoder, VAE can outperform LSTM language models. We demonstrate perplexity gains on two datasets, representing the first positive experimental result on the use VAE for generative modeling of text. Further, we conduct an in-depth investigation of the use of VAE (with our new decoding architecture) for semi-supervised and unsupervised labeling tasks, demonstrating gains over several strong baselines.

Code Implementations3 repos
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes