Research map / Bandits and online learning / Online convex optimization and games
Real-Time Optimisation for Online Learning in Auctions
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use V-Conv-OGA instead of ERM at every step | Online monopoly-price learning for a stationary bidder · Stationary bidder | update complexity: \mathcal{O}(t) -> \mathcal{O}(1) | Helped |
| Use V-Conv-OGA instead of ERM at every step | Online monopoly-price learning for a stationary bidder · Stationary bidder | memory complexity: \mathcal{O}(t) -> \mathcal{O}(1) | Helped |
| Use V-Conv-OGA instead of ERM at every step | Online monopoly-price learning for a stationary bidder · Stationary bidder | convergence rate: \mathcal{O}(1/t) -> \mathcal{O}(1/\sqrt{t}) | Hurt |
| Discretise ERM on a grid of resolution 1/√t | ERM · Stationary bidder | update complexity: \mathcal{O}(t) -> \mathcal{O}(\sqrt{t}) | Helped |
| Discretise ERM on a grid of resolution 1/√t | ERM · Stationary bidder | memory complexity: \mathcal{O}(t) -> \mathcal{O}(\sqrt{t}) | Helped |
| Smooth the instantaneous revenue by convolution | Online gradient ascent for monopoly-price learning · Stationary bidder | gradient bias | Helped |
| Decrease the smoothing kernel over time | Conv-OGA, yielding V-Conv-OGA · Stationary bidder | squared reserve-price error: \tilde{\mathcal{O}}\big(t^{-1/2}\big) | Helped |
| Use a Gaussian kernel that decays slower than 1/√t (ablation) | V-Conv-OGA · Stationary bidder | convergence speed | Hurt |
| Decrease the Gaussian kernel faster than 1/√t (ablation) | V-Conv-OGA · Stationary bidder | convergence stability | Hurt |
| Use Conv-OGA with a constant learning rate and fixed smoothing | Online monopoly-price tracking · Non-stationary bidder | dynamic regret: \mathcal{O}(\sqrt{T}) | Helped |
| Increase the constant learning rate (ablation) | Conv-OGA · Non-stationary bidder | adaptation after a distribution switch | Helped |
| Increase the constant learning rate (ablation) | Conv-OGA · Non-stationary bidder | asymptotic error band | Hurt |
Closest papers by meaning
- Learning to Bid in Repeated First-Price Auctions with Budgets (ICML 2023 · similarity 0.89)
- Learning Bayesian Nash Equilibrium in Auction Games via Approximate Best Response (ICML 2025 · similarity 0.87)
- AuctionNet: A Novel Benchmark for Decision-Making in Large-Scale Games (NeurIPS 2024 · similarity 0.87)
- Increasing Revenue in Efficient Combinatorial Auctions by Learning to Generate Artificial Competition (AAAI 2025 · similarity 0.87)
- No-Regret Online Autobidding Algorithms in First-price Auctions (NeurIPS 2025 · similarity 0.86)
- Learning-Augmented Online Bidding in Stochastic Settings (NeurIPS 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.