Research map / Learning theory and algorithms / Combinatorial and search optimization
Revisiting Implicit Differentiation for Learning Problems in Optimal Control
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 |
|---|---|---|---|
| Use IDOC instead of PDP to compute trajectory derivatives | Equality-constrained LfD · Cartpole, Quadrotor, Robotarm, and Rocket | imitation loss | No clear effect |
| Use IDOC instead of PDP to compute trajectory derivatives | Cartpole equality-constrained LfD · Cartpole | numerical stability: two occasions across the five trials | Helped |
| Use IDOC instead of Safe-PDP for hard inequality constraints | Inequality-constrained LfD · Cartpole, Quadrotor, Robotarm, and Rocket | imitation loss | Helped |
| Use a log-barrier approximation instead of hard inequality constraints | Safe-PDP · Inequality-constrained LfD | imitation loss | Mixed |
| Compute VJPs directly instead of constructing full trajectory derivatives | IDOC · Synthetic benchmark | computation time: by an order of magnitude with model size | Helped |
| Use IDOC instead of PDP for full trajectory derivatives | Synthetic benchmark · Synthetic benchmark | computation time | No clear effect |
| Use IDOC instead of PDP to compute trajectory derivatives | Synthetic benchmark · Synthetic benchmark | numerical stability | Helped |
| Vectorize IDOC computations across timesteps | IDOC · Simulation LfD tasks | computation time: around a 2× speedup over PDP | Helped |
| Compute VJPs directly rather than explicitly constructing trajectory derivatives | IDOC · Simulation LfD tasks | computation time: approx. 10% saving in compute time | Helped |
| Use IDOC instead of PDP for trajectory derivatives | Equality-constrained LfD · Robotarm and Rocket | imitation loss | No clear effect |
Closest papers by meaning
- DiLQR: Differentiable Iterative Linear Quadratic Regulator via Implicit Differentiation (ICML 2025 · similarity 0.85)
- Derivative-Free Policy Optimization for Linear Risk-Sensitive and Robust Control Design: Implicit Regularization and Sample Complexity (NeurIPS 2021 · similarity 0.85)
- Differentiable Nonlinear Model Predictive Control (NeurIPS 2026 · similarity 0.85)
- Constrained Optimization From a Control Perspective via Feedback Linearization (NeurIPS 2025 · similarity 0.85)
- Implicit Bias of Policy Gradient in Linear Quadratic Control: Extrapolation to Unseen Initial States (ICML 2024 · similarity 0.84)
- Neural Time-Reversed Generalized Riccati Equation (AAAI 2024 · 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.