Learning from incomplete data in Bayesian networks with qualitative influences

Andrés R. Masegosa, A. J. Feelders, Linda C. van der Gaag

Abstract: Highlights•Learning Bayesian networks with qualitative influences from incomplete data.•Domain knowledge specified in terms of sign of influences between variables.•Prior knowledge exploited in combination with partially observed data.•Two EM-like procedures are theoretically and empirically evaluated.