# Enhancing Unimodal Latent Representations in Multimodal VAEs through Iterative Amortized Inference

arXiv: https://arxiv.org/abs/2410.11403 · HTML: https://mlsanity.com/paper/2410.11403/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 25 of its 36 references that are in the map.
- First posted on arXiv 15 Oct 2024.

## Builds on (most cited first, 25 of 25)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Hierarchical Text-Conditional Image Generation with CLIP Latents](https://mlsanity.com/paper/2204.06125/) (2022)
- [Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding](https://mlsanity.com/paper/2205.11487/) (2022)
- [Layer Normalization](https://mlsanity.com/paper/1607.06450/) (2016)
- [Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)](https://mlsanity.com/paper/1511.07289/) (2015)
- [Multimodal Machine Learning: A Survey and Taxonomy](https://mlsanity.com/paper/1705.09406/) (2017)
- [Multi-Object Representation Learning with Iterative Variational Inference](https://mlsanity.com/paper/1903.00450/) (2019)
- [MultiMAE: Multi-modal Multi-task Masked Autoencoders](https://mlsanity.com/paper/2204.01678/) (2022)
- [Multimodal Generative Models for Scalable Weakly-Supervised Learning](https://mlsanity.com/paper/1802.05335/) (2018)
- [Variational Mixture-of-Experts Autoencoders for Multi-Modal Deep Generative Models](https://mlsanity.com/paper/1911.03393/) (2019)
- [Learning Factorized Multimodal Representations](https://mlsanity.com/paper/1806.06176/) (2018)
- [Inference Suboptimality in Variational Autoencoders](https://mlsanity.com/paper/1801.03558/) (2018)
- [Iterative Amortized Inference](https://mlsanity.com/paper/1807.09356/) (2018)
- [Joint Multimodal Learning with Deep Generative Models](https://mlsanity.com/paper/1611.01891/) (2016)
- [Generalized Multimodal ELBO](https://mlsanity.com/paper/2105.02470/) (2021)
- [Multimodal Generative Learning Utilizing Jensen-Shannon-Divergence](https://mlsanity.com/paper/2006.08242/) (2020)
- [A survey of multimodal deep generative models](https://mlsanity.com/paper/2207.02127/) (2022)
- [Generative Models of Visually Grounded Imagination](https://mlsanity.com/paper/1705.10762/) (2017)
- [On the Limitations of Multimodal VAEs](https://mlsanity.com/paper/2110.04121/) (2021)
- [Disentangling by Partitioning: A Representation Learning Framework for Multimodal Sensory Data](https://mlsanity.com/paper/1805.11264/) (2018)
- [Multimodal Generative Models for Compositional Representation Learning](https://mlsanity.com/paper/1912.05075/) (2019)
- [Private-Shared Disentangled Multimodal VAE for Learning of Hybrid Latent Representations](https://mlsanity.com/paper/2012.13024/) (2020)

## Abstract (arXiv)

Multimodal variational autoencoders (VAEs) aim to capture shared latent representations by integrating information from different data modalities. A significant challenge is accurately inferring representations from any subset of modalities without training an impractical number (2^M) of inference networks for all possible modality combinations. Mixture-based models simplify this by requiring only as many inference models as there are modalities, aggregating unimodal inferences. However, they suffer from information loss when modalities are missing. Alignment-based VAEs address this by aligning unimodal inference models with a multimodal model through minimizing the Kullback-Leibler (KL) divergence but face issues due to amortization gaps, which compromise inference accuracy. To tackle these problems, we introduce multimodal iterative amortized inference, an iterative refinement mechanism within the multimodal VAE framework. This method overcomes information loss from missing modalities and minimizes the amortization gap by iteratively refining the multimodal inference using all available modalities. By aligning unimodal inference to this refined multimodal posterior, we achieve unimodal inferences that effectively incorporate multimodal information while requiring only unimodal inputs during inference. Experiments on benchmark datasets show that our approach improves inference performance, evidenced by higher linear classification accuracy and competitive cosine similarity, and enhances cross-modal generation, indicated by lower FID scores. This demonstrates that our method enhances inferred representations from unimodal inputs.