Multi-Armed Bandit Exposure Fairness Algorithm
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-armed bandit algorithms prioritize maximizing rewards over ensuring fairness in exposure, leading to biased decision-making and social inequality in applications like recommender systems.
Innovation Solution
A multi-armed bandit method and apparatus that selects K items out of M while ensuring exposure equity, where each item is exposed for at least a certain fraction of time, by using upper confidence bound (UCB) indices and unfairness indices to balance reward maximization with fairness constraints.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional multi-armed bandit algorithms are used to maximize rewards, then the averaged total reward is improved, but exposure fairness of items deteriorates
Solution Approach 1:
The patent modifies the selection criteria by introducing fairness constraints as additional parameters. Instead of solely maximizing reward, the system selects K arms that satisfy both reward optimization and exposure fairness constraints, changing the decision parameters from pure reward-based to constrained optimization parameters.
Solution Approach 2:
The patent implements a dynamic selection mechanism where the set of candidate arms is updated at each time step based on current reward estimates and fairness constraints. The system dynamically adjusts which arms are eligible for selection to balance reward maximization with ensuring minimum exposure thresholds for all arms.
2Productivity
If machine learning algorithms are used for decision-making, then productivity is improved, but bias and social inequality worsen
Solution Approach 1:
The patent incorporates feedback mechanisms that monitor and adjust the selection process to ensure fairness. By tracking the number of times each arm has been selected and comparing it against fairness constraints, the system provides feedback to correct biased decision-making patterns while maintaining high productivity.
Solution Approach 2:
The patent segments the selection process into distinct phases: identifying arms that satisfy fairness constraints and then selecting from those the optimal K arms based on reward. This segmentation separates the fairness enforcement function from the reward optimization function, allowing each to be addressed independently and reducing harmful biases.
Data Source
AI summary
The present disclosure relates to a multi-armed bandit method and apparatus for selecting multiple items while ensuring fairness of exposure of the multiple items and maximizing the averaged total reward. The MAB method includes: initializing the empirical mean reward and number of arm selections of each arm for the M arms, and the time step; incrementing the time step; calculating the UCB index of each arm for the M arms; selecting K−1 arms with the K−1 highest UCB indices calculated; calculating unfairness indices for the unchosen M−(K−1) arms; checking if there is an arm with a positive unfairness index among the unchosen M−(K−1) arms; selecting the remaining single arm depending on whether there is an arm with a positive unfairness index among the unchosen M−(K−1) arms; playing the selected K arms; and updating the empirical mean reward and the number of arm selections for the played arms.