Diffusion Transformers for Tabular Data Time Series Generation
This addresses a largely unexplored domain in tabular data generation for applications requiring time-dependent series, though it appears incremental as it extends an existing framework to new data types.
The paper tackles the problem of generating time series of tabular data, which involves handling heterogeneity and variable lengths, by proposing a Diffusion Transformers (DiTs) based approach. The result shows that this method outperforms previous work by a large margin across six datasets.
Tabular data generation has recently attracted a growing interest due to its different application scenarios. However, generating time series of tabular data, where each element of the series depends on the others, remains a largely unexplored domain. This gap is probably due to the difficulty of jointly solving different problems, the main of which are the heterogeneity of tabular data (a problem common to non-time-dependent approaches) and the variable length of a time series. In this paper, we propose a Diffusion Transformers (DiTs) based approach for tabular data series generation. Inspired by the recent success of DiTs in image and video generation, we extend this framework to deal with heterogeneous data and variable-length sequences. Using extensive experiments on six datasets, we show that the proposed approach outperforms previous work by a large margin.