SlateQ: A Tractable Decomposition for Reinforcement Learning with Recommendation Sets

Authors: Eugene Ie, Vihan Jain, Jing Wang, Sanmit Narvekar, Ritesh Agarwal, Rui Wu, Heng-Tze Cheng, Tushar Chandra, Craig Boutilier

IJCAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 5 Empirical Evaluation: Simulation, 6 Empirical Evaluation: Live Experiments
Researcher Affiliation Collaboration 1Google Research 2Department of Computer Science, University of Texas at Austin
Pseudocode No The paper presents mathematical equations and descriptions of updates, but it does not include formal pseudocode or algorithm blocks.
Open Source Code No The paper mentions that TensorFlow software is available, but it does not provide an explicit statement or link for the open-source code of the methodology described in this paper.
Open Datasets No We construct a simulation environment since most public datasets are point-wise, static, and not designed for evaluating multi-step user-recommender interactions.
Dataset Splits No The paper describes evaluating strategies on 5000 simulated users but does not specify explicit train/validation/test dataset splits with percentages or sample counts.
Hardware Specification No The paper mentions training on 'large-scale recommenders' and using 'distributed training', but it does not provide specific details about the hardware used, such as GPU models, CPU types, or memory.
Software Dependencies No The paper states that 'The model is trained using Tensor Flow', but it does not specify the version number of TensorFlow or other software dependencies.
Experiment Setup Yes The paper specifies parameters for the simulation environment, such as '|T| = 20, m = 10, k = 3', and describes the training approach in live experiments: 'We train on-policy over pairs of consecutive start page visits, with LTV labels computed using Eq. (14), and use top-k optimization for both training and serving'.