Teacher Forcing as Generalized Bayes: Optimization Geometry Mismatch in Switching Surrogates for Chaotic Dynamics
Not on the map. Why?
Abstract (arXiv)
Identity teacher forcing (ITF) enables stable training of deterministic recurrent surrogates for chaotic dynamical systems and has been highly effective for dynamical systems reconstruction (DSR) with recurrent neural networks (RNNs), including interpretable almost-linear RNNs (AL-RNNs). However, as an intervention-based prediction loss (and thus a generalized Bayes update), teacher forcing need not match the free-running model's marginal likelihood geometry. We compare the objective-induced curvatures of ITF and marginal likelihood in a probabilistic switching augmentation of AL-RNNs, estimating ambiguity-aware observed information via Louis' identity. In the switching setting studied here, conditioning on a single forced regime path (as ITF does) inflates curvature, while marginal likelihood curvature is reduced by a missing-information correction when multiple switching explanations remain plausible. In Lorenz-63 experiments, windowed evidence fine-tuning improves held-out evidence but can degrade dynamical quantities of interest (QoIs) relative to ITF-pretrained models.
Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 13 of its 31 references that are in the map. First posted on arXiv 28 Apr 2026.
Builds on
13 of 31 references.
References · 13
- 2019On the Variance of the Adaptive Learning Rate and Beyondcited by 772
- 2013A General Framework for Updating Belief Distributionscited by 175
- 2021On the difficulty of learning chaotic dynamics with RNNscited by 67
- 2023Generalized Teacher Forcing for Learning Chaotic Dynamicscited by 58
- 2013Rao-Blackwellised Particle Filtering for Dynamic Bayesian Networkscited by 47
- 2022Tractable Dendritic RNNs for Reconstructing Nonlinear Dynamical Systemscited by 36
- 2024Out-of-Domain Generalization in Dynamical Systems Reconstructioncited by 33
- 2024DySLIM: Dynamics Stable Learning by Invariant Measure for Chaotic Systemscited by 30
- 2024Almost-Linear RNNs Yield Highly Interpretable Symbolic Codes in Dynamical Systems Reconstructioncited by 19
- 2023Bifurcations and loss jumps in RNN trainingcited by 17
- 2024When are dynamical systems learned from time series data statistically accurate?cited by 15
- 2025Uncovering the Computational Roles of Nonlinearity in Sequence Modeling Using Almost-Linear RNNscited by 5
1 more
- 2025Detecting Invariant Manifolds in ReLU-Based RNNscited by 3
Built on by
Citing papers per year: 2026: 2.
2 papers.