← Search

Adji B. Dieng

2 accepted papers

2019

Avoiding Latent Variable Collapse with Generative Skip Models

AISTATS 2019poster

Variational autoencoders (VAEs) learn distributions of high-dimensional data. They model data with a deep latent-variable model and then fit the model by maximizing a lower bound of the log marginal likelihood. VAEs can capture complex distributions, but they can also suffer from an issue known as "…

Cited by 229SourcePDFScholar
2017

TopicRNN: A Recurrent Neural Network with Long-Range Semantic Dependency

ICLR 2017poster

In this paper, we propose TopicRNN, a recurrent neural network (RNN)-based language model designed to directly capture the global semantic meaning relating words in a document via latent topics. Because of their sequential nature, RNNs are good at capturing the local structure of a word sequence – b…

Cited by 310SourceScholar