Funded Projects

Our group’s research is carried out within the following ongoing funded projects

Society-Aware Machine Learning (SAML)

ERC Starting Grant, European Research Council · 2023–present

SAML develops a “society-aware” approach to machine learning that moves algorithm design away from being driven solely by the interests of technology owners, toward a collaborative process involving owners, users, and regulators alike. The project aims to jointly account for business goals, user benefit, and societal risk at every stage of ML development, from data collection to learning, so that the resulting technologies can be trusted by society. Isabel Valera is the Principal Investigator of this project.

Center for Perspicuous Computing (CPEC)

Transregional Collaborative Research Centre 248, DFG · since 2019, second funding period through 2026

CPEC brings together around 80 researchers across Saarland University, TU Dresden, MPI for Software Systems, and CISPA, with partners at TU Wien and the University of Freiburg. The center lays the scientific foundations for computer systems that can explain their own behavior — “perspicuous” systems — so that humans stay able to understand and control software that participates in decisions affecting them, across a system’s design, operation, and inspection. Isabel Valera is a co-PI within CPEC.

RTG Neuroexplicit Models of Language, Vision, and Action

Research Training Group 2853, DFG · 2023–2028

This DFG-funded Research Training Group trains 24 PhD students under 14 professors across five institutions on the Saarbrücken campus — Saarland University’s Computer Science and Language Science & Technology departments, MPI for Informatics, MPI for Software Systems, CISPA, and DFKI. It develops models that combine neural components with human-interpretable (“explicit,” including neurosymbolic) components for tasks in language processing, vision, and decision-making. Isabel Valera is a co-PI and supervisor within the RTG.

Reasonable AI (RAI)

Cluster of Excellence EXC 3057, German Excellence Strategy (federal & state); building on hessian.AI / 3AI, funded 2021–2025 by the Hessian Ministry of Higher Education, Research, Science and the Arts

RAI is a Cluster of Excellence proposal within hessian.AI aiming at a new generation of AI that learns in more decentralized, continuously-adapting ways, builds abstract knowledge of the world, and can reason and adapt to context — addressing current deep learning’s reliance on huge models, data, and compute, and its weak common-sense reasoning. Work is organized across four research labs (Systemic, Observational, Active, and Challenging AI) and coordinated by Kristian Kersting, Mira Mezini, and Marcus Rohrbach at TU Darmstadt. Isabel Valera is a co-PI, based at Saarland University.