Research map / Architectures and efficiency / RNNs and normalization
Dynamic Game Theoretic Neural Optimizer
Guan-Horng Liu, Tianrong Chen, Evangelos Theodorou
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 |
|---|---|---|---|
| enlarge information structure of baselines from open-loop to closed-loop/cooperative closed-loop (ablation) | SGD, RMSprop, Adam, EKFAC · MNIST, SVHN, CIFAR10/100 residual and inception nets | accuracy | Improved |
| use feedback updates with large learning rates (lr=1.0 MNIST, lr=2.0 SVHN) (ablation) | DGNOpt vs SGD on inception networks · MNIST, SVHN | training stability | Improved |
| increase number of fictitious players N beyond 2 in game-extended SGD (ablation) | SGD on 4-conv 2-FC network · classification, final accuracy and convergence speed | accuracy | Mixed |
| cooperative training with N>1 fictitious agents vs N=1 (ablation) | SGD · feedforward network | convergence speed and minima | Improved |
| adaptive bandit (EXP3++) alignment instead of fixed alignment (ablation) | DGNOpt · CIFAR10 ResNet18 | accuracy: +1% vs baseline, +0.5% vs other strategies | Improved |
| random alignment each iteration instead of fixed alignment (ablation) | DGNOpt · CIFAR10 ResNet18 | convergence speed | Improved |
| replace EMSA with DGNOpt | larger networks · classification datasets | accuracy / stability | Improved |
Closest papers by meaning
- Optimal Control Theoretic Neural Optimizer: From Backpropagation to Dynamic Programming (TPAMI 2025 · similarity 0.91)
- DDPNOpt: Differential Dynamic Programming Neural Optimizer (ICLR 2021 · similarity 0.88)
- Solving Neural Min-Max Games: The Role of Architecture, Initialization & Dynamics (NeurIPS 2025 · similarity 0.88)
- The Mechanics of n-Player Differentiable Games (ICML 2018 · similarity 0.88)
- Recursive Reasoning in Minimax Games: A Level $k$ Gradient Play Method (NeurIPS 2022 · similarity 0.87)
- Learning Equilibria in Adversarial Team Markov Games: A Nonconvex-Hidden-Concave Min-Max Optimization Problem (NeurIPS 2024 · similarity 0.85)
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.
