A Bayesian Networks approach to Operational Risk

摘要: Abstract A system for Operational Risk management based on the computational paradigm of Bayesian Networks is presented. The algorithm allows construction a Network targeted each bank and takes into account in simple realistic way correlations among different processes bank. internal losses are averaged over variable time horizon, so that at times removed, while same kept: thus suitable to perform learning network topology parameters; since main aim understand role losses, assessments domain experts not used. has been validated synthetic series. It should be stressed proposed thought practical implementation mid or small sized bank, it impact organizational structure requires an investment human resources which limited area.