Linear Submodular Bandits with a Knapsack Constraint

Authors: Baosheng Yu, Meng Fang, Dacheng Tao

AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We also conduct a number of experiments and the experimental results confirm our theoretical analyses.
Researcher Affiliation Academia Centre for Quantum Computation and Intelligent Systems, University of Technology, Sydney Department of Computing and Information Systems, The University of Melbourne
Pseudocode Yes Algorithm 1 MCSGreedy Algorithm, Algorithm 2 CGreedy Algorithm
Open Source Code No The paper does not provide concrete access to source code for the methodology described.
Open Datasets Yes following the previous work (Yue and Guestrin 2011), we empirically evaluate our algorithms on simulation dataset by using news article recommendation (Li et al. 2010) as a case study.
Dataset Splits No The paper uses a "simulation dataset" but does not explicitly provide specific dataset split information (percentages, sample counts, or predefined splits) for training, validation, or testing.
Hardware Specification No The paper does not provide specific hardware details used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details with version numbers.
Experiment Setup Yes In this simulation experiment, we randomly generate a d-dimensional vector (x1, x2, . . . , xd) to represent a news article a, where xi (0, 1)(for all i = 1, 2, . . . , d) represents the information coverage of a news article a on the topic i. For each news article a, we assume that it only has a limited number of main topics (xi > 0.5) and noisy topics (xi 0.5). The number of main topics is Nmain = 2 and the number of noise topics is Nnoise = 1. The topics coverage of each article are sampled from a uniform distribution. We randomly generate the w [0, 1]d, which is unknown to our algorithms, to represent a user s interest level in each topic. We also assume that a user will like some topics very much (w (i) 0.8) and will dislike other topics (w (i) 0.1). The costs set C = {c1, c2, . . . , cm} are sampled from the uniform distribution and normal distribution.