Research map / Learning theory and algorithms / Submodular and game algorithms
Kernelized Multiplicative Weights for 0/1-Polyhedral Games: Bridging the Gap Between Learning in Extensive-Form and Normal-Form Games
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 optimistic multiplicative weights with kernelized vertex updates instead of prior optimistic FTRL/OMD | EFG learning · EFGs | per-player regret bound: $\mathcal{O}(\sqrt{m}\log(A)\,\|Q\|_1^2 \ T^{1/4})$ -> $\mathcal{O}(m\log(A)\,\|Q\|_1\ \log^4(T))$ | Helped |
| Use kernelized vertex updates instead of prior non-optimistic MWU-based methods | EFG learning · EFGs | per-player regret bound: $\mathcal{O}(\sqrt{\log A}\,\|Q\|_1 \ T^{1/2})$ -> $\mathcal{O}(\sqrt{\log A}\,\sqrt{\|Q\|_1}\ T^{1/2})$ | Helped |
| Implement OMWU through the sequence-form kernel | KOMWU · Perfect-recall EFGs | per-iteration time complexity: $\mathcal{O}(|\Sigma_i|^2)$ -> $\mathcal{O}(|\Sigma_i|)$ | Helped |
| Use KOMWU with learning rate $\eta=10$ (ablation) | KOMWU · 3-player Leduc poker | cumulative regret | Hurt |
| Use KOMWU rather than CFR or CFR(RM+) | Poker learning under COLS · Kuhn and Leduc poker | maximum per-player cumulative regret | Helped |
Closest papers by meaning
- Efficient Kernelized Learning in Polyhedral Games beyond Full Information: From Colonel Blotto to Congestion Games (NeurIPS 2025 · similarity 0.90)
- From Average-Iterate to Last-Iterate Convergence in Games: A Reduction and Its Applications (NeurIPS 2025 · similarity 0.89)
- Efficient Learning in Polyhedral Games via Best-Response Oracles (AAAI 2024 · similarity 0.89)
- On the Optimality of Dilated Entropy and Lower Bounds for Online Learning in Extensive-Form Games (NeurIPS 2024 · similarity 0.89)
- Fast Last-Iterate Convergence of Learning in Games Requires Forgetful Algorithms (NeurIPS 2024 · similarity 0.89)
- Extensive-Form Game Solving via Blackwell Approachability on Treeplexes (NeurIPS 2024 · similarity 0.88)
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.