Relational Boosted Bandits

Authors: Ashutosh Kakadiya, Sriraam Natarajan, Balaraman Ravindran12123-12130

AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We empirically demonstrate the effectiveness and interpretability of RB2 on tasks such as link prediction, relational classification, and recommendation., Empirical Evaluation Our experimental evaluation explicitly aims to answer the following specific questions: 1. Effectiveness: How does RB2 perform against other baselines on real-world data?, Figure 1 presents the cumulative regret of RB2 with other algorithms on synthetic and real-world data sets.
Researcher Affiliation Academia Ashutosh Kakadiya,1 Sriraam Natarajan, 2 Balaraman Ravindran 1 1 Robert Bosch Centre for Data Science and Artificial Intelligence, Indian Institute of Technology Madras 2 The University of Texas at Dallas
Pseudocode Yes Algorithm 1 Boosted Relational Bandits: Softmax Exploration with Informed Sampling, Algorithm 2 Informed Sampling: Stochastic Prioritization,, Algorithm 3 Greedy Sampling
Open Source Code No The paper provides pseudocode, but does not explicitly state that source code is provided or offer a link to an open-source repository.
Open Datasets Yes Simulated Movie data set, Movie Lense data set (Motl and Schulte 2015), Drug-Drug Interaction(DDI) (Dhami et al. 2018), ICML Co-author (Dhami et al. 2020), IMDB (Mihalkova and Mooney 2007), sports data of Never Ending Language Learner (NELL) data set (Mitchell et al. 2018)
Dataset Splits No The paper refers to 'train', 'validation', and 'test' in the JSON schema context, but does not explicitly describe train/validation/test dataset splits (e.g., percentages, counts, or predefined splits) for reproducibility of experiments. It refers to 'mini-batches' for online learning and 'batch training' but not standard data splits.
Hardware Specification No The paper does not provide specific details about the hardware used to run the experiments.
Software Dependencies No The paper does not provide specific ancillary software details with version numbers (e.g., library or solver names with version numbers like Python 3.8, CPLEX 12.4).
Experiment Setup Yes The hyperparameter description that has been employed in our experiments are present in Table 2. In Table 2: Batch Size, Trees per batch (K). and Typically, in our experiments, the number of trees (K) is preset to 4 K 10.