Constrained Efficient Global Optimization of Expensive Black-box Functions

Authors: Wenjie Xu, Yuning Jiang, Bratislav Svetozarevic, Colin Jones

ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Numerical experiments on sampled instances from the Gaussian process, artificial numerical problems, and a black-box building controller tuning problem all demonstrate the competitive performance of our algorithm.
Researcher Affiliation Academia 1Automatic Control Laboratory, EPFL, Lausanne, Switzerland 2Urban Energy Systems Laboratory, Empa, Zurich, Switzerland.
Pseudocode Yes Algorithm 1 Lower confidence bounds based CONstrained ef FIcient Global (CONFIG) optimization algorithm.
Open Source Code No The paper states: 'The experiments are implemented in python, based on the package GPy (GPy, since 2012).' This indicates the use of a third-party package but does not state that the authors' own code for the described methodology is open-source or provide a link to it.
Open Datasets No The paper uses sampled instances from Gaussian processes, artificial numerical instances constructed from given functions (Tab. 4), and data generated from a building simulator (Energym). It does not use or provide access information for any publicly available or open datasets in the traditional sense.
Dataset Splits No The paper does not provide specific training/validation/test dataset splits. The experiments involve sampled or simulated data, rather than pre-existing datasets with established splits.
Hardware Specification No The paper states: 'In our experiments, all the problems have low-dimensional input ( 3). So we use pure grid search to solve the auxiliary problem for different algorithms and therefore, the computational time is almost the same for different algorithms.' This implies general computing resources but does not provide specific hardware details like GPU/CPU models, processor types, or memory.
Software Dependencies No The paper states: 'The experiments are implemented in python, based on the package GPy (GPy, since 2012).' It mentions the software 'GPy' and its founding year but does not specify a precise version number for GPy or Python, nor does it list other software dependencies with versions.
Experiment Setup Yes The paper states: 'manually setting βi,t = 3 works well. We also set λ = 0.052 as the noise variance for the Gaussian process modelling. We use the common squared exponential kernel functions. For Sec. 6.2 and Sec. 6.3, we randomly sample a few points and maximize the likelihood function to get the hyperparameters of the kernel.'