Bernoulli Flow Models: Self-Consistent Generative Modeling for Binary Data
To address this fundamental limitation and decouple the generative dynamics from fixed discrete time steps, we propose Bernoulli Flow Models (BFM).
Key points
- Binary diffusion models typically require a large number of function evaluations (NFEs) to generate high-quality samples, making practical inference computationally expensive.
- Existing binary diffusion models define a discrete one-step forward path and then derive the reverse posterior.
- Rather than relying on sequential one-step Markov diffusion chains, BFM defines a unified continuous global Bernoulli probability flow path between data distributions and pure noise, from which we derive analytical closed-form posterior transitions over arbitrary time intervals.
- These results establish BFM as a theoretically rigorous, self-consistent, and practically effective framework for fast binary data generation.
Sources (1)
- [1]Bernoulli Flow Models: Self-Consistent Generative Modeling for Binary DataarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 8, 06:52 AM
To address this fundamental limitation and decouple the generative dynamics from fixed discrete time steps, we propose Bernoulli Flow Models (BFM).
Binary diffusion models typically require a large number of function evaluations (NFEs) to generate high-quality samples, making practical inference computationally expensive.
Extractive summary: sentences quoted from the sources.