Randomized Wagering Mechanisms

Authors: Yiling Chen, Yang Liu, Juntao Wang1845-1852

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

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we conduct simulations to evaluate the average-case performance of wagering mechanisms. We show that LWS and RP-SWME are more efficient than existing deterministic (weakly) incentive compatible mechanisms WSWM, NAWM and DCA. Meanwhile, we show that RP-SWME has smaller payoff variance and higher probability of winning money than LWS. In the simulations, we generate the predictions of agents using two models: i). the logit model (Satop a a et al. 2014), ii). the synthetic model (Ranjan and Gneiting 2010; Allard, Comunian, and Renard 2012; Satop a a et al. 2014).
Researcher Affiliation Academia Yiling Chen,1 Yang Liu,2 Juntao Wang1 1Harvard University, 2University of California, Santa Cruz yiling@seas.harvard.edu, yangliu@ucsc.edu, juntaowang@g.harvard.edu
Pseudocode Yes Mechanism 1 Lottery Wagering Mechanisms, Mechanism 2 Surrogate Wagering Mechanisms, Algorithm 3 Error Rate Selection Algorithm, Mechanism 4 Random Partition SWME (RP-SWME)
Open Source Code No The paper does not provide an explicit statement about open-source code availability or a link to a code repository for the described methodology.
Open Datasets No The paper describes generating predictions using models like logit and synthetic, and wagers using uniform and Pareto distributions, but it does not specify or provide access information for a public dataset in the traditional sense.
Dataset Splits No The paper describes simulation setups and data generation methods but does not provide specific train/validation/test dataset splits or mention a validation set for reproducibility.
Hardware Specification No The paper describes conducting simulations but does not provide any specific hardware details such as GPU models, CPU types, or memory specifications used for these experiments.
Software Dependencies No The paper does not provide specific software dependencies or version numbers needed to replicate the experiment, beyond mentioning the use of the Brier score.
Experiment Setup Yes In the simulations, we generate the predictions of agents using two models: i). the logit model (Satop a a et al. 2014), ii). the synthetic model (Ranjan and Gneiting 2010; Allard, Comunian, and Renard 2012; Satop a a et al. 2014). The details of these models can be found in Section 8.1 of the full version (Chen, Liu, and Wang 2018). We generate the wagers of agents by two distributions: i). a uniform distribu- tion over [0,1], ii). a Pareto distribution with shape parameter 1.16 and scale parameter 1, characterizing the 20% of the population has 80% of the wealth (Freeman, Pennock, and Vaughan 2017). We use Brier score as the scoring rule used in the wagering mechanisms that we evaluate.