CBRAP: Contextual Bandits with RAndom Projection

Authors: Xiaotian Yu, Michael R. Lyu, Irwin King

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

Reproducibility Variable Result LLM Response
Research Type Experimental By comparing with three benchmark algorithms, we demonstrate improved performance on cumulative payoffs of CBRAP during its sequential decisions on both synthetic and real-world datasets, as well as its superior time-efficiency. ... We evaluate the CBRAP algorithm via a series of experiments with synthetic and real-world datasets.
Researcher Affiliation Academia Xiaotian Yu, Michael R. Lyu, Irwin King Department of Computer Science and Engineering The Chinese University of Hong Kong, Shatin, N.T., Hong Kong Email: {xtyu,lyu,king}@cse.cuhk.edu.hk
Pseudocode Yes Algorithm 1 CBRAP
Open Source Code Yes Our algorithm and used datasets are all publicly available1 1https://github.com/Aaronyxt/CBRAP
Open Datasets Yes Then, we conduct experiments on two real-world datasets, i.e., Movielens2 and Jester3. ... 2http://grouplens.org/datasets/movielens/ 3http://www.ieor.berkeley.edu/ goldberg/jester-data/
Dataset Splits No The paper describes the datasets used and the performance metric (cumulative payoffs over T=1000 rounds) but does not provide specific details on training, validation, or test dataset splits.
Hardware Specification Yes We conduct all experiments on a server installed with Ubuntu 12.04.5 LTS, which contains 24 processors of each core being Intel CPU@2.60GHz, and has a total memory of 200GB.
Software Dependencies No The paper mentions 'Ubuntu 12.04.5 LTS' as the operating system, but does not provide specific version numbers for any other software dependencies, libraries, or frameworks used in the experiments.
Experiment Setup No The paper mentions inputs like 'm, T, β R+ and α R+' for Algorithm 1 and explores different values of 'm' in experiments (m = 10, 20, 30, 40, 50), but it does not specify concrete hyperparameters like learning rates, batch sizes, or other training configurations.