Research map / Learning theory and algorithms / Submodular and game algorithms
Probably Approximately Efficient Combinatorial Auctions via Machine Learning
Gianluca Brero, Benjamin Lubin, Sven Seuken
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 |
|---|---|---|---|
| add SVR generalization of bids instead of XOR bids | One-shot Poly-2 auction · GSVM Full-Random q=40 | efficiency: 43% -> 92% | Improved |
| add SVR generalization of bids instead of XOR bids | One-shot Exp auction · LSVM Pos-Random q=40 | efficiency: 75% -> 82% | Improved |
| add iterative querying of allocated bundles | Exp-kernel one-shot auction · LSVM Pos-Random q=40 | efficiency: 82% -> 92% | Improved |
| add iterative querying of allocated bundles | Gaussian-kernel one-shot auction · GSVM Pos-Random q=40 | efficiency: 86% -> 87% | No clear effect |
| use Exp kernel instead of Gaussian kernel | One-shot SVR auction · LSVM Full-Random q=40 | efficiency: 51% -> 70% | Improved |
| use Poly-2 kernel instead of Gaussian kernel | One-shot SVR auction · GSVM Full-Random q=40 | efficiency: 83% -> 92% | Improved |
| sample from bidder's bundles of interest (Pos-Random) instead of full space | One-shot Exp auction · GSVM q=40 | efficiency: 86% -> 99% | Improved |
| increase number of initial queries q from 10 to 40 | One-shot Exp auction · GSVM Full-Random | efficiency: 57% -> 86% | Improved |
| increase insensitivity threshold zeta from 0.1 to 10 (ablation) | Iterative Exp auction · LSVM Full-Random q=40 | runtime: 484.1 -> 53.5 | Improved |
| increase insensitivity threshold zeta from 0.1 to 10 (ablation) | Iterative Exp auction · LSVM Full-Random q=40 | efficiency: 77% -> 74% | Hurt |
| increase insensitivity threshold zeta from 0.1 to 10 (ablation) | Iterative Gaus auction · LSVM Full-Random q=40 | efficiency: 64% -> 69% | Improved |
| tune hyperparameters on the wrong value model (ablation) | Iterative Exp auction · GSVM Full-Random q=10 | efficiency: -6% | Hurt |
Closest papers by meaning
- Prices, Bids, Values: One ML-Powered Combinatorial Auction to Rule Them All (ICML 2025 · similarity 0.93)
- Machine Learning-Powered Combinatorial Clock Auction (AAAI 2024 · similarity 0.91)
- Increasing Revenue in Efficient Combinatorial Auctions by Learning to Generate Artificial Competition (AAAI 2025 · similarity 0.88)
- Learning Optimal Auctions with Correlated Value Distributions (AAAI 2025 · similarity 0.87)
- Learning Bayesian Nash Equilibrium in Auction Games via Approximate Best Response (ICML 2025 · similarity 0.86)
- Predicting Real-World Penny Auction Durations by Integrating Game Theory and Machine Learning (AAAI 2024 · 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.
