Julia Herbinger
Papers
1
Total Citations
4
H-Index
1
About
Julia Herbinger is a rising researcher at the intersection of explainable artificial intelligence and Bayesian optimization, with a focus on making black-box optimization transparent and collaborative. Her most-cited work, "Explaining Bayesian Optimization by Shapley Values Facilitates Human-AI Collaboration" (2024, 4 citations), addresses a critical paradox: while Bayesian optimization (BO) with Gaussian processes is a cornerstone algorithm for optimizing expensive, unknown functions, it remains opaque to human users. Herbinger’s key contribution is the application of Shapley values—a game-theoretic concept from explainable AI—to decompose BO’s parameter proposals, revealing why specific inputs are suggested. This innovation transforms BO from a “black box” into a tool for human-AI collaboration, enabling domain experts to trust, verify, and refine optimization decisions. Though early in her career, her work bridges a significant gap between algorithmic efficiency and interpretability, with potential applications in engineering design, drug discovery, and automated machine learning. By pioneering explainable BO, Herbinger is laying the groundwork for more transparent, accountable optimization systems that empower researchers to harness AI’s power without sacrificing understanding.
Research Focus
Key Achievements
Top Papers
- 1