Exact Learning of Preference Structure: Single-peaked Preferences and Beyond

Authors: Sonja Kraiczy, Edith Elkind

ICML 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental To test the dependence of the sample size on p, and to see how well the lower bound in Proposition 3.9 performs, we ran experiments with m = 1000 alternatives (we choose a relatively large value of m for robustness; as argued above the results are likely to be very similar for other values of m), varying p as p {0.001, 0.002, . . . , 0.5}. Figure 1 shows the average sample size over 1000 repetitions needed to identify the underlying axis from Up( ) depending on p(1 p) (red), together with a plot of p(1 p) (blue), in log-log scale.
Researcher Affiliation Academia 1Department of Computer Science, University of Oxford, Oxford, United Kingdom. Correspondence to: Edith Elkind <edith.elkind@ox.ac.uk>.
Pseudocode No No structured pseudocode or algorithm blocks were found in the paper.
Open Source Code No The paper does not provide any statement or link indicating that the source code for the described methodology is publicly available.
Open Datasets No The paper describes experiments involving sampling from theoretical distributions to verify bounds, not training models on traditional datasets or providing access to such datasets.
Dataset Splits No The paper does not specify any training/validation/test dataset splits. The experiments are simulations to verify theoretical bounds on sample size.
Hardware Specification No No specific hardware details (e.g., GPU/CPU models, memory) used for running the experiments are provided in the paper.
Software Dependencies No The paper does not provide specific names or version numbers for any software dependencies used in the experiments.
Experiment Setup Yes To test the dependence of the sample size on p, and to see how well the lower bound in Proposition 3.9 performs, we ran experiments with m = 1000 alternatives (we choose a relatively large value of m for robustness; as argued above the results are likely to be very similar for other values of m), varying p as p {0.001, 0.002, . . . , 0.5}. Figure 1 shows the average sample size over 1000 repetitions needed to identify the underlying axis from Up( ) depending on p(1 p) (red), together with a plot of p(1 p) (blue), in log-log scale.