Deep generative models as the probability transformation functions
This work provides a foundational theoretical framework for generative modeling, potentially benefiting researchers and practitioners in machine learning by enabling more efficient and effective techniques.
The paper tackles the problem of unifying diverse deep generative models by proposing a theoretical perspective that views them as probability transformation functions, demonstrating that all such models transform simple distributions into complex data distributions, which facilitates methodological transfer and universal theoretical approaches.
This paper introduces a unified theoretical perspective that views deep generative models as probability transformation functions. Despite the apparent differences in architecture and training methodologies among various types of generative models - autoencoders, autoregressive models, generative adversarial networks, normalizing flows, diffusion models, and flow matching - we demonstrate that they all fundamentally operate by transforming simple predefined distributions into complex target data distributions. This unifying perspective facilitates the transfer of methodological improvements between model architectures and provides a foundation for developing universal theoretical approaches, potentially leading to more efficient and effective generative modeling techniques.