Understanding and removing bias for a fair and explainable AIдоклад на конференции