On Kernelized Multi-armed Bandits

Authors: Sayak Ray Chowdhury, Aditya Gopalan

ICML 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, experimental evaluation and comparisons to existing algorithms on synthetic and real-world environments are carried out that highlight the favorable gains of the proposed strategies in many cases.
Researcher Affiliation Academia Department of Electrical Communication Engineering, Indian Institute of Science, Bengaluru, 560012, India.
Pseudocode Yes Algorithm 1 Improved-GP-UCB (IGP-UCB) and Algorithm 2 GP-Thompson-Sampling (GP-TS) are provided with structured pseudocode.
Open Source Code No The paper does not provide a direct link to open-source code for the described methodology or state that the code is publicly available.
Open Datasets Yes We use temperature data collected from 54 sensors deployed in the Intel Berkeley Research lab [...] http://db.csail.mit.edu/labdata/labdata. html. We take light sensor data collected in the CMU Intelligent Workplace in Nov 2005, which is available online as Matlab structure http://www.cs.cmu.edu/~guestrin/Class/ 10708-F08/projects/lightsensor.zip
Dataset Splits No The paper mentions running experiments for 30000 iterations over 25 independent trials, and using initial days of sensor data to 'learn algorithm parameters', and subsequent days for 'testing'. However, it does not explicitly provide specific train/validation/test splits, percentages, or cross-validation details for the datasets.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running its experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers (e.g., Python, PyTorch, TensorFlow versions or specific solver versions).
Experiment Setup Yes We set noise parameter R2 to be 1% of function range and use λ = R2. We used Squared Exponential kernel with lengthscale parameter l = 0.2 and Mat ern kernel with parameters ν = 2.5, l = 0.2. Parameters βt, βt, vt of IGP-UCB, GP-UCB and GP-TS are chosen as given in Section 3, with δ = 0.1, B2 = f T Kf and γt set according to theoretical upper bounds for corresponding kernels.