Unity by Diversity: Improved Representation Learning in Multimodal VAEs
This work addresses representation learning for multimodal data analysis, offering an incremental improvement over existing multimodal VAE methods.
The paper tackles the problem of learning shared representations in multimodal variational autoencoders by replacing hard architectural constraints with a soft mixture-of-experts prior, resulting in improved latent representations and better imputation of missing data modalities across benchmark and real-world datasets.
Variational Autoencoders for multimodal data hold promise for many tasks in data analysis, such as representation learning, conditional generation, and imputation. Current architectures either share the encoder output, decoder input, or both across modalities to learn a shared representation. Such architectures impose hard constraints on the model. In this work, we show that a better latent representation can be obtained by replacing these hard constraints with a soft constraint. We propose a new mixture-of-experts prior, softly guiding each modality's latent representation towards a shared aggregate posterior. This approach results in a superior latent representation and allows each encoding to preserve information better from its uncompressed original features. In extensive experiments on multiple benchmark datasets and two challenging real-world datasets, we show improved learned latent representations and imputation of missing data modalities compared to existing methods.