Vanishing gradients from GP initialization explain HDBO failures; MLE of length scales suffices for SOTA performance, enabling the MSR variant on real-world tasks.
Cylindrical Thompson Sampling for High-Dimensional Bayesian Optimization
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Understanding High-Dimensional Bayesian Optimization
Vanishing gradients from GP initialization explain HDBO failures; MLE of length scales suffices for SOTA performance, enabling the MSR variant on real-world tasks.