Interpretable tabular data generation

Abstract Generative adversarial network () models have been successfully utilized in a wide range of machine learning applications, and tabular data generation domain is not an exception. Notably, some state-of-the-art generation, such as , etc. are based on models. Even though these resulted superior performance generating artificial when trained datasets, there lot room (and desire) for improvement. Not to mention that existing methods do weaknesses other than performance. For example, the current focus only model, limited emphasis given interpretation model. Secondly, operate raw features only, hence they fail exploit any prior knowledge explicit feature interactions can be during process. To alleviate two above-mentioned limitations, this work, we propose novel model— G enerative A dversarial Network modelling inspired from N aive B ayes L ogistic R egression’s relationship ( $${ { \texttt {GANBLR} } }$$ GANBLR ), which address limitation -based but provides capability handle well. Through extensive evaluations demonstrate ’s well better interpretable (explanation importance synthetic process) compared