Evidential Sparsification of Multimodal Latent Spaces in Conditional Variational Autoencoders
This addresses computational challenges in multimodal AI applications like motion planning and behavior prediction, but it is incremental as it builds on existing latent space methods.
The paper tackles the problem of high computational cost in downstream tasks due to large discrete latent spaces in conditional variational autoencoders by proposing a post hoc sparsification technique using evidential theory to reduce the latent space size while preserving multimodality, with experiments on image generation and human behavior prediction showing effectiveness.
Discrete latent spaces in variational autoencoders have been shown to effectively capture the data distribution for many real-world problems such as natural language understanding, human intent prediction, and visual scene representation. However, discrete latent spaces need to be sufficiently large to capture the complexities of real-world data, rendering downstream tasks computationally challenging. For instance, performing motion planning in a high-dimensional latent representation of the environment could be intractable. We consider the problem of sparsifying the discrete latent space of a trained conditional variational autoencoder, while preserving its learned multimodality. As a post hoc latent space reduction technique, we use evidential theory to identify the latent classes that receive direct evidence from a particular input condition and filter out those that do not. Experiments on diverse tasks, such as image generation and human behavior prediction, demonstrate the effectiveness of our proposed technique at reducing the discrete latent sample space size of a model while maintaining its learned multimodality.