Research map / Learning theory and algorithms / Sample complexity
Refined bounds for algorithm configuration: The knife-edge of dual class approximability
Maria-Florina Balcan, Tuomas Sandholm, Ellen Vitercik
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 |
|---|---|---|---|
| Approximate dual performance functions under L∞ using a simpler function class | Algorithm configuration generalization bounds · General algorithm configuration | generalization guarantee | Improved |
| Rely on finite-Lp approximation instead of L∞ approximation | Algorithm configuration generalization bounds · Worst case | generalization guarantee | Hurt |
| Use L∞ dual-function approximation with structural risk minimization | CPLEX variable-selection policy configuration · CATS arbitrary, score_L/score_S | sample complexity for generalization guarantee 0.1: 10^5 versus 7 \cdot 10^7; 700 times | Improved |
| Choose approximation complexity by structural risk minimization rather than fixing it to the maximum number of observed pieces (ablation) | CPLEX variable-selection policy configuration · CATS integer programming experiments | sample complexity: up to 4.5 times smaller | Improved |
| Use L∞ approximation with structural risk minimization rather than the maximum-piece baseline (ablation) | CPLEX variable-selection policy configuration · CATS arbitrary, score_L/score_S | samples for generalization error 0.25: 6700 samples to 1500 samples; 4.6x improvement | Improved |
| Use L∞ approximation with structural risk minimization rather than the maximum-piece baseline (ablation) | CPLEX variable-selection policy configuration · CATS integer programming experiments | samples for generalization error 0.3: 4300 samples to 1400 samples; 3.07x improvement | Improved |
Closest papers by meaning
- Learning-Augmented Approximation Algorithms for Maximum Cut and Related Problems (NeurIPS 2024 · similarity 0.86)
- Certification-Enhanced Generalization Bounds (NeurIPS 2026 · similarity 0.86)
- The Sample Complexity of Gradient Descent in Stochastic Convex Optimization (NeurIPS 2024 · similarity 0.84)
- Algorithmic Stability Unleashed: Generalization Bounds with Unbounded Losses (ICML 2024 · similarity 0.84)
- Sample complexity of stochastic optimization with integer variables (NeurIPS 2026 · similarity 0.84)
- Slower Generalization, Faster Memorization: A Sweet Spot in Algorithmic Learning (NeurIPS 2026 · similarity 0.84)
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.
