Algorithmic Bias in AI
This session would pose a particular focus on the potential harmful consequences generated by the injection of the human-biases into AI, which lead to the so-called “Algorithmic Bias”. The main objectives of the session are: 1) describe the role of algorithmic bias embedded in machine learning applications (e.g. recommender systems, rankings, NLP, computer vision, etc..); 2) present methods able to analyze and mitigate those biases, with an emphasis on recommender systems and rankings.
The session will be covering the following topics:
introduction of algorithmic bias in AI (e.g. algorithmic unfairness, polarization)
auditing algorithms for detecting harmful consequences of this bias, with an emphasis on recommender systems
mitigation strategies to reduce unfairness in recommendation and ranking systems.
#DataWeek2022