Learning Valuation Distributions from Partial Observation

Authors: Avrim Blum, Yishay Mansour, Jame Morgenstern

AAAI 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical Our main result in this work is to solve this problem efficiently. Namely, we derive a polynomial time algorithm (with polynomial sample complexity) that recovers an approximation b Di of each of the distributions Di, down to some price pγ, below which there is at most γ probability of any bidder winning.
Researcher Affiliation Collaboration Computer Science Department, Carnegie Mellon University avrim@cs.cmu.edu Yishay Mansour Tel Aviv University Microsoft Research, Hertzelia mansour@tau.ac.il Jamie Morgenstern Computer Science Department Carnegie Mellon University jamiemmt@cs.cmu.edu
Pseudocode Yes Algorithm Kaplan: Estimates the CDF of i from samples with reserves; Algorithm IWin: Est. P[i wins in [ℓτ, ℓτ+1]| maxj bj < ℓτ+1]; Algorithm Intervals: Partitions bid space to est. fi
Open Source Code No The paper does not provide any statement or link regarding the availability of open-source code.
Open Datasets No The paper is theoretical and does not refer to specific public datasets used for empirical training. It discusses a theoretical distribution 'P over items' and drawing 'x at random from P'.
Dataset Splits No The paper is theoretical and does not provide specific training/validation/test dataset splits. It discusses 'm samples' in a theoretical context.
Hardware Specification No The paper is theoretical and does not mention any hardware specifications used for experiments.
Software Dependencies No The paper is theoretical and does not list any software dependencies with version numbers.
Experiment Setup No The paper is theoretical and does not provide details about an experimental setup, hyperparameters, or training settings.