Fair Allocation based on Diminishing Differences

Authors: Erel Segal-Halevi, Haris Aziz, Avinatan Hassidim

IJCAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Simulations based on a simple random model show that with high probability, a necessarily-proportional allocation does not exist but a necessarily-DDproportional allocation exists. Moreover, that allocation is proportional according to the underlying cardinal utilities.
Researcher Affiliation Academia Erel Segal-Halevi Bar-Ilan University erelsgl@gmail.com Haris Aziz Data61 CSIRO and UNSW haris.aziz@data61.csiro.au Avinatan Hassidim Bar-Ilan University avinatan@cs.biu.ac.il
Pseudocode Yes Repeat as long as there are items: For i = 1 . . . n: give agent i his best remaining item. For i = n . . . 1: give agent i his best remaining item.
Open Source Code No The paper does not provide any explicit statement about releasing source code or a link to a code repository.
Open Datasets No The paper states: "To simulate such rankings, we determined for each item a market value drawn uniformly at random from [1, 2]. We determined the value of each item to each agent as the item s market value plus noise drawn uniformly at random from [ A, A], where A is a parameter.". This indicates custom-generated data for simulations, not a publicly available dataset with concrete access information.
Dataset Splits No The paper uses randomly generated utility profiles for simulations but does not specify any train/validation/test splits, fold cross-validation, or specific random seeds for data partitioning, as it's a simulation study.
Hardware Specification No The paper mentions running "Simulations" but does not specify any hardware used (e.g., CPU, GPU models, memory, etc.).
Software Dependencies No The paper mentions running "Simulations" but does not specify any software versions (e.g., programming languages, libraries, or specific solvers with versions).
Experiment Setup No The paper describes the random generation of utility profiles for simulations: "We determined for each item a market value drawn uniformly at random from [1, 2]. We determined the value of each item to each agent as the item s market value plus noise drawn uniformly at random from [ A, A], where A is a parameter." It also mentions repeating the experiment "1000 times for different values of A {0.1, . . . , 1} and for different numbers of items 2m items for m {2, . . . , 8}". This provides some setup parameters, but lacks common hyperparameters or system-level training settings typically found in experimental setups for machine learning models.