LGCVOct 7, 2020

Conditional Generative Modeling via Learning the Latent Space

arXiv:2010.03132v210 citations
Originality Incremental advance
AI Analysis

This addresses the need for multimodal generation in machine learning, offering a general-purpose solution that can replace engineered pipelines, though it appears incremental as it builds on existing latent variable methods.

The paper tackles the problem of deterministic inference limiting single-modal applications in deep learning by proposing a conditional generative framework using latent variables for multimodal spaces, resulting in faster convergence, better representations, and outperforming domain-specific pipelines on various tasks.

Although deep learning has achieved appealing results on several machine learning tasks, most of the models are deterministic at inference, limiting their application to single-modal settings. We propose a novel general-purpose framework for conditional generation in multimodal spaces, that uses latent variables to model generalizable learning patterns while minimizing a family of regression cost functions. At inference, the latent variables are optimized to find optimal solutions corresponding to multiple output modes. Compared to existing generative solutions, in multimodal spaces, our approach demonstrates faster and stable convergence, and can learn better representations for downstream tasks. Importantly, it provides a simple generic model that can beat highly engineered pipelines tailored using domain expertise on a variety of tasks, while generating diverse outputs. Our codes will be released.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes