Efficient and Robust High-Dimensional Linear Contextual Bandits

Authors: Cheng Chen, Luo Luo, Weinan Zhang, Yong Yu, Yijiang Lian

IJCAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results demonstrate the effectiveness of our algorithm and verify our theoretical guarantees.
Researcher Affiliation Collaboration Cheng Chen1 , Luo Luo2 , Weinan Zhang1 , Yong Yu1 and Yijiang Lian3 1Shanghai Jiao Tong University 2Hong Kong University of Science and Technology 3Baidu jack chen1990@sjtu.edu.cn, luoluo@ust.hk, {wnzhang, yyu}@apex.sjtu.edu.cn, lianyijiang@baidu.com
Pseudocode Yes Algorithm 1 SCFD and Algorithm 2 CBSCFD
Open Source Code No The paper states 'The code is implemented in Matlab R2017b.' but does not provide any links to its source code or explicitly state its public release.
Open Datasets Yes We perform our experiments on two real-world data sets: MNIST [Le Cun et al., 1998] and CIFAR10 [Krizhevsky and Hinton, 2009].
Dataset Splits No The paper describes how data samples are drawn for online classification in a sequential manner for the contextual bandit problem, rather than specifying traditional train/validation/test dataset splits with percentages or counts.
Hardware Specification No We conduct all experiments on a Linux server which contains 8 processors and has total memory of 32GB. (This lacks specific CPU/GPU models, only providing general system characteristics).
Software Dependencies Yes The code is implemented in Matlab R2017b.
Experiment Setup Yes The parameter β of all methods is searched in {10 4, 10 3, . . . , 1} and λ is searched in {2 10 4, 2 10 3, . . . , 2 104}. We choose the best values for each approach and report the average results in Figure 1.