Invited Talks
Dealing with bias and fairness in data science social good projects
Tackling issues of bias and fairness in AI has received increased attention from the research community in recent years, yet a lot of the research has focused on theoretical aspects and very little extensive empirical work has been done on real-world policy problems. Treating bias and fairness as primary metrics of interest, from project scopint to model building and selection should be a standard practice in data science for social good (DSSG) projects. In this talk we will try to bridget the gap between research and practice, by deep diving into algorithmic fairness, from metrics and definitions to practical case studies in DSSG, including bias audits using the Aequitas toolkit (http://github.com/dssg/aequitas).
Short bio:
Pedro Saleiro is a senior research manager at Feedzai Research where he leads the FATE (Fairness, Accountability, Transparency, and Ethics) research group. He is responsible for several initiatives related to improving model explainability in the context of financial crime prevention, bias auditing and algorithmic fairness, experimentation and A/B testing, ML governance and reproducibility. Previously, Pedro was postdoc at the University of Chicago and research data scientist at the Center for Data Science and Public Policy working with Rayid Ghani, developing new methods and open source tools, and doing data science for social good projects with government and non-profit partners in diverse policy areas. Pedro was a data science mentor at the Data Science for Social Good Summer Fellowship 2018.
Data for Good by Design: Concrete Examples
In recent years, compelling cases have shown the value of using privately-held data and data science to social good. However, the move beyond ad-hoc projects into sustainable initiatives has proved challenging. This talk starts with an overview of examples of data collaboratives of private-sector data sharing for social good involving large organizations (e.g. UNICEF, Telefonica, Bloomberg, Microsoft). The talk then delves into practical examples of how data for good can be done by design from the start. It showcases the role of market solutions in creating win-win business models and maximizing social impact, while upholding the highest standards of data protection and privacy.
Short bio:
Natalia Adler is the Co-Founder and CEO of Pebble Analytics, a social impact startup that uses data science, technology and market solutions to forecast social risks, Natalia Adler brings over 14 years of experience leveraging innovative solutions for social problems at the United Nations Children's Fund (UNICEF). Natalia conceptualized and ran data collaboratives with the private sector aimed at using data science to tackle the world's most complex challenges, including epidemics, suicides, urban mobility, migration and forced displacement. She has introduced a Human Centered Design approach to support policymaking in Nicaragua, developed a Sustainability Framework for Latin America; fostered the creation of 'entrepreneurial ecosystems’ in Central America, and supported equitable and pro-poor spending through Public Finance Management analytics in Mozambique. Natalia holds a Master’s in Human Rights from Columbia University and a Bachelor’s in French Literature from the University of Pennsylvania, where she graduated summa cum laude and with the highest honors.