Research map / Learning theory and algorithms / Combinatorial and search optimization
Solving Stackelberg Prediction Game with Least Squares Loss via Spherically Constrained Least Squares Reformulation
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 |
|---|---|---|---|
| Reformulate SPG-LS as spherically constrained least squares and solve with Krylov (LTRSR) method instead of SOCP | SPG-LS · Synthetic dense and sparse data | running time: up to ~26786x faster on sparse data; ~11-32x faster on dense data | Helped |
| Use RTRNewton on the SCLS sphere instead of SOCP | SPG-LS · Synthetic dense data | running time: roughly 2x faster than SOCP at largest dense size, but slower than SDP on red wine | Helped |
| Use LTRSR instead of SOCP on real-world datasets | SPG-LS · Red wine, residential building, insurance, blog | running time: LTRSR fastest among all four methods on all real datasets | Helped |
| Replace SOCP with SCLS (LTRSR) solver | SPG-LS · Real datasets (wine, building) | relative objective error vs SOCP: errors up to 3.37e-5 | No clear effect |
| Use gamma=0.01 instead of 0.1 on dense synthetic data (ablation) | SCLS LTRSR vs SOCP · Synthetic dense, gamma=0.01 | running time: LTRSR under 6 s vs SOCP up to ~110 s; similar speedup pattern | No clear effect |
| Increase dimension/sparsity (sparser, larger n) for LTRSR vs SOCP (ablation) | SPG-LS · Sparse synthetic | speedup ratio: ratio grows from ~318 (sparsity 0.01, small size) to ~26786 (sparsity 0.0001, large size) | Helped |
Closest papers by meaning
- Error Analysis of Spherically Constrained Least Squares Reformulation in Solving the Stackelberg Prediction Game (NeurIPS 2024 · similarity 0.96)
- Fast Algorithms for Stackelberg Prediction Game with Least Squares Loss (ICML 2021 · similarity 0.93)
- Calibrated Stackelberg Games: Learning Optimal Commitments Against Calibrated Agents (NeurIPS 2023 · similarity 0.86)
- Learning in Stackelberg Mean Field Games: A Non-Asymptotic Analysis (NeurIPS 2025 · similarity 0.85)
- Prediction-Intervention Games and Invariant Sets (NeurIPS 2026 · similarity 0.84)
- Prediction Accuracy of Learning in Games : Follow-the-Regularized-Leader meets Heisenberg (ICML 2024 · similarity 0.83)
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.