Risk-Sensitive Diffusion: Robustly Optimizing Diffusion Models with Noisy Samples
This addresses robust generation for noisy data in real-world applications like tabular and time-series domains, but it is incremental as it adapts existing diffusion methods to a new setting.
The paper tackles the problem of diffusion models degrading on noisy non-image data by introducing a risk-sensitive SDE parameterized by data quality vectors, and it shows significant performance improvements over baselines on tabular and time-series datasets.
Diffusion models are mainly studied on image data. However, non-image data (e.g., tabular data) are also prevalent in real applications and tend to be noisy due to some inevitable factors in the stage of data collection, degrading the generation quality of diffusion models. In this paper, we consider a novel problem setting where every collected sample is paired with a vector indicating the data quality: risk vector. This setting applies to many scenarios involving noisy data and we propose risk-sensitive SDE, a type of stochastic differential equation (SDE) parameterized by the risk vector, to address it. With some proper coefficients, risk-sensitive SDE can minimize the negative effect of noisy samples on the optimization of diffusion models. We conduct systematic studies for both Gaussian and non-Gaussian noise distributions, providing analytical forms of risk-sensitive SDE. To verify the effectiveness of our method, we have conducted extensive experiments on multiple tabular and time-series datasets, showing that risk-sensitive SDE permits a robust optimization of diffusion models with noisy samples and significantly outperforms previous baselines.