Local Differential Privacy for Bayesian Optimization

Authors: Xingyu Zhou, Jian Tan11152-11159

AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We further conduct empirical comparisons of different algorithms over both synthetic and real-world datasets, which demonstrate the superior performance of our new Mo MA-GP-UCB algorithm in both private and non-private settings.
Researcher Affiliation Collaboration Xingyu Zhou,1 Jian Tan 2 1 ECE, The Ohio State University 2 Alibaba Group, Sunnyvale
Pseudocode Yes Algorithm 1 LDP-ATA-GP-UCB; Algorithm 2 LDP-TGP-UCB; Algorithm 3 Mo MA-GP-UCB
Open Source Code No The paper does not provide any explicit statements about releasing source code or links to a code repository.
Open Datasets Yes Light sensor data. This data is collected in the CMU Intelligent Workplace in Nov 2005, which is available online as Matlab structure4; 4http://www.cs.cmu.edu/∼guestrin/Class/10708-F08/projects
Dataset Splits No The paper mentions "601 train samples and 192 test samples" for the Light sensor data but does not explicitly state the use of a validation set or specific train/validation/test splits for any of the datasets.
Hardware Specification No The paper does not specify any hardware details (e.g., CPU, GPU models, memory) used for running the experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers (e.g., Python version, library versions).
Experiment Setup Yes The parameters used for each algorithm are set order-wise similar to those recommended by the theorems. We run each algorithm for 10 independent trials... The parameters for the kernel function are l = 0.2 for k SE and l = 0.2, ν = 2.5 for k Mat ern. We set B = maxx D |f(x)| and y(x) = f(x) + η. For the LDP case, the noise η is uniformly sampled in [−1, 1] and hence R = 1. For the non-private heavy-tailed case, the noise η are samples from the Student’s t-distribution with 3 degrees of freedom. Hence, v = B2 + 3 and c = 3.