RAYEN: Imposition of Hard Convex Constraints on Neural Networks
cs.LG, cs.RO
Submitted: 2023-07-17
Updated: 2026-11-05
Code: https://github.com/leggedrobotics/rayen
Terminology
Sources
- Differentiating Through a Cone Program
- PDP: A General Neural Framework for Learning Constraint Satisfaction Solvers
- mPOLICE: Provable Enforcement of Multi-Region Affine Constraints in Deep Neural Networks
- POLICEd RL: Learning Closed-Loop Robot Control Policies with Provable Satisfaction of Hard Constraints
- Improving Feasibility via Fast Autoencoder-Based Projections
- DC3: A learning method for optimization with hard constraints
- Eigenvalue and Generalized Eigenvalue Problems: Tutorial
- Deep Declarative Networks: A New Hope
- Pinet: Optimizing hard-constrained neural networks with orthogonal projection layers
- Linearly Constrained Neural Networks
- Not Only Rewards But Also Constraints: Applications on Legged Robot Locomotion
- Reinforcement Learning for Versatile, Dynamic, and Robust Bipedal Locomotion Control
- Imposing Hard Constraints on Deep Networks: Promises and Limitations
- HardNet: Hard-Constrained Neural Networks with Universal Approximation Guarantees
- FSNet: Feasibility-Seeking Neural Network for Constrained Optimization with Guarantees
- Soft-Radial Projection for Constrained End-to-End Learning
- Actuator-Constrained Reinforcement Learning for High-Speed Quadrupedal Locomotion
- LMI-Net: Linear Matrix Inequality--Constrained Neural Networks via Differentiable Projection Layers
- Generating Large Convex Polytopes Directly on Point Clouds
Related papers
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis
- (Mis)generalization of Helpful-Only Fine-tuning