Research map / Learning theory and algorithms / Submodular and game algorithms
PreferenceNet: Encoding Human Preferences in Auction Design with Deep Learning
Neehar Peri, Michael Curry, Samuel Dooley, John P Dickerson
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Learn the quota constraint from exemplars rather than optimize it explicitly | PreferenceNet vs RegretNet · 4x2 unit-demand auction | PCA: .1 -> 100.0 | Improved |
| Learn the quota constraint from exemplars rather than optimize it explicitly | PreferenceNet vs RegretNet · 2x2 additive auction | PCA: 75.1 -> 100.0 | Improved |
| Use PreferenceNet to enforce quota | PreferenceNet · 4x4 additive auction | PCA: 0 -> 0 | No clear effect |
| Learn preferences from exemplars instead of explicitly optimizing the known constraint | PreferenceNet vs RegretNet · Synthetic fairness auctions | PCA: 2x2 unit-demand TVF: 100.0 -> 100.0; 4x2 unit-demand quota: .1 -> 100.0; 4x4 additive quota: 0 -> 0 | Mixed |
| Add label noise following the probit model (ablation) | PreferenceNet · 2-agent, 2-item auctions; synthetic preference labels | PCA: 25% | Hurt |
| Add label noise following the probit model (ablation) | PreferenceNet · 2-agent, 2-item auctions; synthetic preference labels | mean regret: 25% | No clear effect |
| Enforce preference loss with augmented Lagrangian multipliers (ablation) | PreferenceNet · Synthetic fairness auctions | mean regret | Hurt |
| Enforce preference loss with augmented Lagrangian multipliers (ablation) | PreferenceNet · Synthetic fairness auctions | mean payments | Hurt |
| Enforce preference loss with augmented Lagrangian multipliers (ablation) | PreferenceNet · Synthetic fairness auctions | PCA: 4x4 additive quota: 0 -> .1 | Mixed |
| Add preference MLP training and co-training | PreferenceNet · TVF auctions, 1x2 additive | training time: 28m -> 43m | Hurt |
| Train on mixtures of TVF, entropy, and quota preference labels (ablation) | PreferenceNet · Additive auctions, mixed synthetic preferences | average PCA: 78.8; 99.9; 16.0; 64.4 | Mixed |
| Use PreferenceNet rather than RegretNet under asymmetric valuations | 2x2 auction · 2x2 asymmetric valuations, quota | PCA: 39.2 -> 100.0; 27.3 -> 100.0; 17.1 -> 100.0 | Improved |
Closest papers by meaning
- AuctionNet: A Novel Benchmark for Decision-Making in Large-Scale Games (NeurIPS 2024 · similarity 0.90)
- Auctionformer: A Unified Deep Learning Algorithm for Solving Equilibrium Strategies in Auction Games (ICML 2024 · similarity 0.89)
- BundleFlow: Deep Menus for Combinatorial Auctions by Diffusion-Based Optimization (NeurIPS 2025 · similarity 0.89)
- Learning Optimal Auctions with Correlated Value Distributions (AAAI 2025 · similarity 0.89)
- Predicting Real-World Penny Auction Durations by Integrating Game Theory and Machine Learning (AAAI 2024 · similarity 0.87)
- Increasing Revenue in Efficient Combinatorial Auctions by Learning to Generate Artificial Competition (AAAI 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
