Multi-Source Neural Variational Inference
This work addresses the challenge of integrating complementary and redundant information from multiple sources for machine learning applications, presenting an incremental improvement over existing variational inference methods.
The paper tackles the problem of learning from multiple information sources by proposing a variational autoencoder framework with separate encoders per source, enabling the computation of divergence measures between source-specific posterior approximations. It demonstrates the method's effectiveness in learning shared representations and structured output prediction, and shows how conflict detection and redundancy can improve inference robustness.
Learning from multiple sources of information is an important problem in machine-learning research. The key challenges are learning representations and formulating inference methods that take into account the complementarity and redundancy of various information sources. In this paper we formulate a variational autoencoder based multi-source learning framework in which each encoder is conditioned on a different information source. This allows us to relate the sources via the shared latent variables by computing divergence measures between individual source's posterior approximations. We explore a variety of options to learn these encoders and to integrate the beliefs they compute into a consistent posterior approximation. We visualise learned beliefs on a toy dataset and evaluate our methods for learning shared representations and structured output prediction, showing trade-offs of learning separate encoders for each information source. Furthermore, we demonstrate how conflict detection and redundancy can increase robustness of inference in a multi-source setting.