An adversarially trained autoencoder learns a convex latent space to enable rapid approximate projections that enforce nonconvex constraints in optimization and reinforcement learning.
FSNet: Feasibility-seeking neural network for constrained optimization with guarantees.arXiv preprint arXiv:2506.00362
6 Pith papers cite this work. Polarity classification is still indexing.
6
Pith papers citing it
representative citing papers
SnareNet introduces a repair layer that navigates the range space of constraints plus adaptive relaxation training to enforce hard non-convex constraints on neural network outputs more reliably than prior methods.
Action aliasing from safety projections harms policy-gradient estimates more severely when the projection is inside the policy than when it is outside, but a penalty term restores competitiveness.
A transformer predicts unit commitment schedules over 72 hours, heuristics fix infeasibilities, and the output warm-starts a MILP solver to achieve 100% feasibility, faster runtimes, and lower costs in 20% of single-bus test cases.
A survey synthesizing stochastic, robust, and distributionally robust optimization methods for energy infrastructure planning under uncertainty while identifying gaps and machine learning opportunities.
citing papers explorer
-
Improving Feasibility via Fast Autoencoder-Based Projections
An adversarially trained autoencoder learns a convex latent space to enable rapid approximate projections that enforce nonconvex constraints in optimization and reinforcement learning.
-
SnareNet: Flexible Repair Layers for Neural Networks with Hard Constraints
SnareNet introduces a repair layer that navigates the range space of constraints plus adaptive relaxation training to enforce hard non-convex constraints on neural network outputs more reliably than prior methods.
-
Safe Reinforcement Learning using Action Projection: Safeguard the Policy or the Environment?
Action aliasing from safety projections harms policy-gradient estimates more severely when the projection is inside the policy than when it is outside, but a penalty term restores competitiveness.
-
A Multi-Stage Warm-Start Deep Learning Framework for Unit Commitment
A transformer predicts unit commitment schedules over 72 hours, heuristics fix infeasibilities, and the output warm-starts a MILP solver to achieve 100% feasibility, faster runtimes, and lower costs in 20% of single-bus test cases.
-
Optimization Under Uncertainty for Energy Infrastructure Planning: A Synthesis of Methods, Tools, and Open Challenges
A survey synthesizing stochastic, robust, and distributionally robust optimization methods for energy infrastructure planning under uncertainty while identifying gaps and machine learning opportunities.
- HardNet++: Nonlinear Constraint Enforcement in Neural Networks