Eytan Bakshy

About

Eytan Bakshy is a researcher specializing in Bayesian optimization and scalable machine learning methods for complex, real-world optimization problems. His work addresses some of the most pressing challenges in the field, particularly the difficulty of applying Bayesian optimization effectively in high-dimensional and multi-output settings. In his highly influential 2020 paper, "Re-Examining Linear Embeddings for High-Dimensional Bayesian Optimization" (32 citations), Bakshy and his collaborators critically revisited and challenged prevailing assumptions about linear embedding techniques, offering a more rigorous framework for scaling Bayesian optimization to high-dimensional parameter spaces without sacrificing sample efficiency. Building on this foundation, his 2021 work, "Bayesian Optimization with High-Dimensional Outputs" (9 citations), extended the methodology to handle objectives defined over many correlated outcomes, addressing practical scientific and engineering scenarios where multiple interdependent metrics must be optimized simultaneously. Together, these contributions reflect a research agenda focused on making principled probabilistic optimization techniques more broadly applicable and computationally tractable. Bakshy's work is particularly valuable for practitioners and researchers seeking robust, sample-efficient solutions to large-scale optimization challenges in science and industry.

Research Focus

Key Achievements

Top Papers

1. 1
   Re-Examining Linear Embeddings for High-Dimensional Bayesian Optimization

   32 citations · 2020
2. 2
   Bayesian Optimization with High-Dimensional Outputs

   9 citations · 2021

Key Collaborators

Contact & Links