Neural Topic Modeling with Cycle-Consistent Adversarial Training
This work addresses the problem of improving topic coherence and document inference in neural topic modeling for researchers in natural language processing, representing an incremental advancement over prior adversarial methods.
The authors tackled the limitations of existing neural topic models by proposing ToMCAT and sToMCAT, which use cycle-consistent adversarial training to infer topic distributions for documents and incorporate labels for better coherence, outperforming competitive baselines in topic modeling and text classification.
Advances on deep generative models have attracted significant research interest in neural topic modeling. The recently proposed Adversarial-neural Topic Model models topics with an adversarially trained generator network and employs Dirichlet prior to capture the semantic patterns in latent topics. It is effective in discovering coherent topics but unable to infer topic distributions for given documents or utilize available document labels. To overcome such limitations, we propose Topic Modeling with Cycle-consistent Adversarial Training (ToMCAT) and its supervised version sToMCAT. ToMCAT employs a generator network to interpret topics and an encoder network to infer document topics. Adversarial training and cycle-consistent constraints are used to encourage the generator and the encoder to produce realistic samples that coordinate with each other. sToMCAT extends ToMCAT by incorporating document labels into the topic modeling process to help discover more coherent topics. The effectiveness of the proposed models is evaluated on unsupervised/supervised topic modeling and text classification. The experimental results show that our models can produce both coherent and informative topics, outperforming a number of competitive baselines.