DeepVoting: Learning Voting Rules with Tailored Embeddings

2024-08-24

arXiv.org

AI Summary

This work addresses multi-agent preference aggregation, seeking voting rules that simultaneously satisfy formal axiomatic guarantees and admit efficient learnability. Method: We propose a probabilistic social choice function framework, modeling voting rules as mappings from preference distributions to candidate ranking distributions. We introduce the first voting-targeted preference embedding method, significantly enhancing learning efficiency and scalability. Guided by social choice constraints—particularly a strengthened variant of the No-Show paradox immunity—we incorporate axiomatic requirements directly into the learning process via fine-tuning. Contribution/Results: Leveraging deep neural networks, our approach accurately reproduces classical voting rules and maintains high accuracy even under large-scale electorates. It is the first learnable voting mechanism that is (i) axiomatically verifiable, (ii) structurally interpretable, and (iii) scalable in training—thereby bridging rigorous social choice theory with modern machine learning for preference aggregation.

Technology Category

• Machine Learning: Learning Preferences or Rankings
• Game Theory and Economic Paradigms: Social Choice / Voting
• Multiagent Systems: Mechanism Design

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Social networks and social learning
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Aggregating the preferences of multiple agents into a collective decision is a common step in many important problems across areas of computer science including information retrieval, reinforcement learning, and recommender systems. As Social Choice Theory has shown, the problem of designing algorithms for aggregation rules with specific properties (axioms) can be difficult, or provably impossible in some cases. Instead of designing algorithms by hand, one can learn aggregation rules, particularly voting rules, from data. However, the prior work in this area has required extremely large models, or been limited by the choice of preference representation, i.e., embedding. We recast the problem of designing a good voting rule into one of learning probabilistic versions of voting rules that output distributions over a set of candidates. Specifically, we use neural networks to learn probabilistic social choice functions from the literature. We show that embeddings of preference profiles derived from the social choice literature allows us to learn existing voting rules more efficiently and scale to larger populations of voters more easily than other work if the embedding is tailored to the learning objective. Moreover, we show that rules learned using embeddings can be tweaked to create novel voting rules with improved axiomatic properties. Namely, we show that existing voting rules require only minor modification to combat a probabilistic version of the No Show Paradox.

Problem

• Learning probabilistic voting rules from data
• Improving efficiency with canonical preference embeddings
• Fine-tuning rules to resist strategic manipulation

Innovation

• Neural networks learn probabilistic social choice functions
• Canonical embeddings improve learning efficiency and network size
• Fine-tuning with axiomatic properties resists specific voting attacks