Wasserstein Distributionally Robust Inverse Multiobjective Optimization

Authors: Chaosheng Dong, Bo Zeng5914-5921

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

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, we demonstrate the effectiveness of our method on both a synthetic multiobjective quadratic program and a real world portfolio optimization problem. Experiments In this section, we provide an MQP and a portfolio optimization problem to illustrate the performance of Algorithm 1.
Researcher Affiliation Collaboration Chaosheng Dong1*, Bo Zeng2 1Amazon 2University of Pittsburgh chaosd@amazon.com, bzeng@pitt.edu
Pseudocode Yes Algorithm 1 Wasserstein Distributionally Robust IMOP
Open Source Code No The paper states 'All the algorithms are programmed with Julia (Bezanson et al. 2017)' but does not provide a link or explicit statement about the availability of the source code for the described methodology.
Open Datasets No The paper describes how synthetic data and real-world case study data are generated/derived ('We first compute Pareto optimal solutions {xi}i [N] by solving WP with weight samples {wi}i [N] that are uniformly chosen from W2. Next, the noisy decision yi is obtained by adding noise to xi for each i [N].' and 'The dataset is derived from monthly total returns of 30 stocks from a blue-chip index...'). It refers to 'supplementary material' for 'true expected returns and true return covariance matrix', but does not provide a concrete link or citation to an openly accessible public dataset with author attribution or a repository for direct download.
Dataset Splits Yes Here, we use an independent validation set that consists of 105 noisy decisions generated in the same way as the training data to compute the prediction error.
Hardware Specification No No specific hardware details (e.g., GPU models, CPU types, or memory) are provided.
Software Dependencies No The paper mentions 'Gurobi' and 'Julia' but does not specify their version numbers. It also refers to 'Baron (Sahinidis 1996)' without a version.
Experiment Setup Yes K = 6 weights from W2 are evenly sampled. The radius ϵ of the Wasserstein ambiguity set is selected from the set {10 4, 10 3, 10 2, 10 1, 1}. The stopping criteria δ is set to be 0.1.