Professor, University of California, Santa Cruz
Luca de Alfaro's research interestes center on fairness in AI, and on computational ecology.
Generally, AI systems are characterized in terms of their average performance. Professor de Alfaro and his co-authros developed DivExplorer, a tool that can quickly analyze black-box systems, and identify the subgroups of data that are treated in anomalous fashion. This enables, for instance, the identification of groups that are most discriminated by admissions, hiring, parole, and similar schemes. The work started on structured data, and is now being extended to non-structured data such as speech and images, and to remediation in addition to detection.
In parallel, Professor de Alfaro is working on computational ecology, and specifically to GPU-based methods for the computation of habitat connectivity and conservation planning.
Previous research interests of Professor de Alfaro included formal methods, game theory, crowdsourcing, and reputation systems.