Efficient Robust Bayesian Optimization for Arbitrary Uncertain inputs

Authors: Lin Yang, Junlong Lyu, Wenlong Lyu, Zhitang Chen

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

Reproducibility Variable Result LLM Response
Research Type Experimental Comprehensive evaluations on synthetic functions and real problems in Sec.5 demonstrate that our algorithm can efficiently identify robust optimum under complex input uncertainty and achieve a state-of-the-art performance. and 5 Evaluation In this section, we first experimentally demonstrate AIRBO s ability to model uncertain inputs of arbitrary distributions, then validate the Nyström-based inference acceleration for GP posterior, followed by experiments on robust optimization of synthetic functions and real-world benchmark.
Researcher Affiliation Industry Lin Yang Huawei Noah s Ark Lab China yanglin33@huawei.com Junlong Lyu Huawei Noah s Ark Lab Hong Kong SAR, China lyujunlong@huawei.com Wenlong Lyu Huawei Noah s Ark Lab China lvwenlong2@huawei.com Zhitang Chen Huawei Noah s Ark Lab Hong Kong SAR, China chenzhitang2@huawei.com
Pseudocode No No explicit pseudocode or algorithm blocks were found. The methods are described in narrative text.
Open Source Code Yes 1The code will be available on https://github.com/huawei-noah/HEBO, and more implementation details can be found in Appendix C.1.
Open Datasets Yes Comprehensive evaluations on synthetic functions and real problems in Sec.5 demonstrate that our algorithm can efficiently identify robust optimum under complex input uncertainty and achieve a state-of-the-art performance. and To evaluate AIRBO in a real-world problem, we employ a robust robot pushing benchmark from [31]
Dataset Splits No The paper refers to 'training datasets' (e.g., 'produce training datasets of D = {(xi, f(xi + δi))|δi Pxi}10 i=1') but does not provide specific training/validation/test split percentages or sample counts for any dataset used.
Hardware Specification No The paper mentions 'GPU memory' and 'parallel computation' but does not provide specific hardware details such as GPU/CPU models, memory specifications, or types of computing instances used for experiments.
Software Dependencies No In our implementation of AIRBO, we design the kernel k used for MMD estimation to be a linear combination of multiple Rational Quadratic kernels as its long tail behavior circumvents the fast decay issue of kernel [6]. We implement our algorithm 1 based on Bo Torch [2] and employ a linear combination of multiple rational quadratic kernels [6] to compute the MMD as Eq. 9. No version numbers for Bo Torch or other software.
Experiment Setup Yes In our implementation of AIRBO... we employ a classic UCB-based acquisition as Eq. 5 with β = 2.0 and maximize it via an L-BFGS-B optimizer.