Implicit Dynamical Flow Fusion (IDFF) for Generative Modeling
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2024
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866916760728371200 |
|---|---|
| author | Rezaei, Mohammad R. Popovic, Milos R. Lankarany, Milad Krishnan, Rahul G. |
| author_facet | Rezaei, Mohammad R. Popovic, Milos R. Lankarany, Milad Krishnan, Rahul G. |
| contents | Conditional Flow Matching (CFM) models can generate high-quality samples from a non-informative prior, but they can be slow, often needing hundreds of network evaluations (NFE). To address this, we propose Implicit Dynamical Flow Fusion (IDFF); IDFF learns a new vector field with an additional momentum term that enables taking longer steps during sample generation while maintaining the fidelity of the generated distribution. Consequently, IDFFs reduce the NFEs by a factor of ten (relative to CFMs) without sacrificing sample quality, enabling rapid sampling and efficient handling of image and time-series data generation tasks. We evaluate IDFF on standard benchmarks such as CIFAR-10 and CelebA for image generation, where we achieve likelihood and quality performance comparable to CFMs and diffusion-based models with fewer NFEs. IDFF also shows superior performance on time-series datasets modeling, including molecular simulation and sea surface temperature (SST) datasets, highlighting its versatility and effectiveness across different domains.\href{https://github.com/MrRezaeiUofT/IDFF}{Github Repository} |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2409_14599 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Implicit Dynamical Flow Fusion (IDFF) for Generative Modeling Rezaei, Mohammad R. Popovic, Milos R. Lankarany, Milad Krishnan, Rahul G. Machine Learning Conditional Flow Matching (CFM) models can generate high-quality samples from a non-informative prior, but they can be slow, often needing hundreds of network evaluations (NFE). To address this, we propose Implicit Dynamical Flow Fusion (IDFF); IDFF learns a new vector field with an additional momentum term that enables taking longer steps during sample generation while maintaining the fidelity of the generated distribution. Consequently, IDFFs reduce the NFEs by a factor of ten (relative to CFMs) without sacrificing sample quality, enabling rapid sampling and efficient handling of image and time-series data generation tasks. We evaluate IDFF on standard benchmarks such as CIFAR-10 and CelebA for image generation, where we achieve likelihood and quality performance comparable to CFMs and diffusion-based models with fewer NFEs. IDFF also shows superior performance on time-series datasets modeling, including molecular simulation and sea surface temperature (SST) datasets, highlighting its versatility and effectiveness across different domains.\href{https://github.com/MrRezaeiUofT/IDFF}{Github Repository} |
| title | Implicit Dynamical Flow Fusion (IDFF) for Generative Modeling |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2409.14599 |