Industry Best Practices in Algorithmic Bias Detection and Mitigation
No ratings
Presented at GlobalPrivacySummit 2019 by
Algorithmic decision-making processes can perpetuate existing biases or create new biases, generate harmful categorizations, or unfairly discriminate against protected classes or specific cohorts. However, if biases are accounted for in their development, algorithms can also generate fairness in decision-making. This panel will present the findings from a multidisciplinary project and research paper that included insights from industry and academic experts interested in algorithmic bias detection and mitigation. Panel experts will share the results of the final paper, which include a formal bias impact assessment and template and a series of policy recommendations such as regulatory sandboxes, safe harbors and consumer algorithmic literacy. Panel experts will also debate the role of government and other stakeholders, including businesses, technologists and consumers, around algorithmic fairness and accountability.What you’ll take away:Understand the ways that biases can be introduced into the process of creating, developing and using an algorithmUnderstand the tools that can be used to detect biasDiscover potential ways to mitigate bias and the pros and cons of existing methodologies