Safe Exploration for Optimization with Gaussian Processes

Authors: Yanan Sui, Alkis Gotovos, Joel Burdick, Andreas Krause

ICML 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate SAFEOPT on synthetic data, as well as two real applications: movie recommendation, and therapeutic spinal cord stimulation.
Researcher Affiliation Academia Yanan Sui YSUI@CALTECH.EDU California Institute of Technology, Pasadena, CA, USA Alkis Gotovos ALKISG@INF.ETHZ.CH ETH Zurich, Zurich, Switzerland Joel W. Burdick JWB@ROBOTICS.CALTECH.EDU California Institute of Technology, Pasadena, CA, USA Andreas Krause KRAUSEA@ETHZ.CH ETH Zurich, Zurich, Switzerland
Pseudocode Yes Algorithm 1 SAFEOPT
Open Source Code No The paper does not provide any specific links or statements about the availability of the source code for the SAFEOPT methodology.
Open Datasets Yes We test the algorithms on the Movie Lens-100k dataset
Dataset Splits No The paper does not provide specific training, validation, or test dataset splits (e.g., percentages, sample counts, or references to predefined splits) to reproduce the partitioning of data for model training and evaluation.
Hardware Specification No The paper does not specify any hardware details (e.g., GPU/CPU models, memory, or specific computer specifications) used for running the experiments.
Software Dependencies No The paper mentions applying 'matrix factorization' and using a 'squared exponential ARD kernel' but does not specify any software names with version numbers (e.g., Python, PyTorch, scikit-learn, or specific solvers).
Experiment Setup Yes The safety threshold is set equal to the mean of all ratings. We apply a matrix factorization with k = 20 latent factors. We ran both SAFE-UCB and SAFEOPT for T = 100 iterations.