Scalable Kernel Inverse Optimization

Authors: Youyuan Long, Tolga Ok, Pedro Zattoni Scroccaro, Peyman Mohajerin Mohajerin Esfahani

NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, we validate the generalization capabilities of the proposed KIO model and the effectiveness of the SSO algorithm through learning-from-demonstration tasks on the Mu Jo Co benchmark.
Researcher Affiliation Academia Youyuan Long Delft Center for Systems and Control Delft University of Technology The Netherlands longyouyuan432@gmail.comTolga Ok Delft Center for Systems and Control Delft University of Technology The Netherlands T.Ok@tudelft.nlPedro Zattoni Scroccaro Delft Center for Systems and Control Delft University of Technology The Netherlands P.Zattoni Scroccaro@tudelft.nlPeyman Mohajerin Esfahani Delft Center for Systems and Control Delft University of Technology The Netherlands P.Mohajerin Esfahani@tudelft.nl
Pseudocode Yes Algorithm 1 Sequential Selection Optimization (SSO)
Open Source Code Yes To foster reproducibility and further research, we provide an opensource implementation of the proposed KIO model and the SSO algorithm, along with the source code of the experiments in Github1. 1https://github.com/Longyouyuan/Scalable-Kernel-Inverse-Optimization
Open Datasets Yes KIO is implemented in its simplified version (9), incorporating a Gaussian kernel, and tested on continuous control datasets from the D4RL benchmark [18]. [18] Justin Fu, Aviral Kumar, Ofir Nachum, George Tucker, and Sergey Levine. D4rl: Datasets for deep data-driven reinforcement learning, 2020.
Dataset Splits No The paper states that the model is 'trained using the SSO Algorithm 1' and 'assessed over 100 test episodes,' but it does not specify explicit training/validation/test dataset splits (e.g., percentages or sample counts) or cite a specific standard split methodology for D4RL.
Hardware Specification No The paper mentions that solving certain problems 'requires up to 256GB of memory,' indicating a memory requirement, but it does not specify the actual hardware used such as GPU/CPU models, types, or speeds.
Software Dependencies Yes The paper mentions using 'CVXPY [13]' and 'off-the-shelf solvers, such as MOSEK [3].' MOSEK is cited with a version number: 'Mosek Ap S. Mosek optimization toolbox for matlab. User s Guide and Reference Manual, Version, 4:1, 2019.'
Experiment Setup Yes All hyperparameters used in this experiment for KIO are listed in Appendix B.