Constrained Optimization (Distance Minimization Objective) on Two Moons 1,500 problems (test)
100FeasibilityFAB: 2 Decoder Nets
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| FAB: 2 Decoder NetsType=ML-Based2026.04 | 100 | 0.5 | 5.64 | |
| FAB: 3 Decoder NetsType=ML-Based2026.04 | 100 | 0.6 | 5.44 | |
| FSNetType=ML-Based2026.04 | 99.3 | 4.03 | 5.3 | |
| FABType=ML-Based2026.04 | 88.3 | 0.41 | 4.22 | |
| Interior PointType=Classical2026.04 | 79.9 | 42 | 6.12 | |
| Projected GradientType=Classical2026.04 | 79.2 | 69.79 | 4.65 | |
| Homeomorphic ProjectionType=ML-Based2026.04 | 76.1 | 243 | 8.12 | |
| FAB: Phase 1 OnlyType=ML-Based2026.04 | 64.1 | 0.54 | 3.83 | |
| Penalty NNType=ML-Based2026.04 | 40.7 | 0.24 | 4.34 | |
| FAB: No Hinge LossType=ML-Based2026.04 | 28.1 | 0.56 | 5.14 | |
| Augmented LagrangianType=Classical2026.04 | 15.7 | 42.58 | 6.43 | |
| Penalty MethodType=Classical2026.04 | 14.6 | 38.67 | 6.39 | |
| FAB: No DξType=ML-Based2026.04 | 13.6 | 0.55 | 3.61 |