mlsanityData to 14 Oct 2026

Retracing Hodgkin and Huxley: State Recovery Does Not Certify Mechanism

arXiv 2609.34566 ↗First posted 28 Sep 2026cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Predicting observed dynamics does not establish recovery of the underlying physical mechanism. Can machine learning retrace the hidden-state reasoning behind the Hodgkin-Huxley (HH) model? We train structured latent models on simulated current and voltage, withholding gate identities and trajectories from training and model selection. We then test response prediction, state recovery, protocol transfer, and agreement with HH dynamics. Prediction error and its cross-seed spread both drop sharply at three latent dimensions under the tested protocols, while gate recovery under new protocols improves through five to six coordinates. State recovery depends on which observations the chart uses. Observed voltage improves current-clamp decoding relative to freely predicted voltage. Under voltage clamp, adding latent state to command voltage raises m-state R2 from 0.976 to above 0.99, yet the transported field disagrees with HH on identical smooth samples. Known invertible HH coordinates achieve high fast-m field agreement under the same audit procedure. An exact HH identity decomposes the discrepancy into time-scale-weighted state error and a residual in the transported field; these terms can cancel or reinforce. These findings concern the tested models and charts. They support evaluating state and dynamics recovery separately, including chart inputs and transported-field agreement across interventions.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 22 of its 46 references that are in the map. First posted on arXiv 28 Sep 2026.

Builds on

22 of 46 references.

References · 22

  1. 2018Neural Ordinary Differential Equationscited by 3,779
  2. 2020Neural Controlled Differential Equations for Irregular Time Seriescited by 506
  3. 2017Deep learning for universal linear embeddings of nonlinear dynamicscited by 494
  4. 2022On the Parameterization and Initialization of Diagonal State Space Modelscited by 424
  5. 2020Universal Differential Equations for Scientific Machine Learningcited by 353
  6. 2015Deep Kalman Filterscited by 166
  7. 2017Deep Learning for Physical Processes: Incorporating Prior Scientific Knowledgecited by 96
  8. 2022Discovering Governing Equations from Partial Measurements with Deep Delay Autoencoderscited by 58
  9. 2019Learning Dynamical Systems from Partial Observationscited by 36
  10. 2021Discovering Sparse Interpretable Dynamics from Partial Observationscited by 25
  11. 2019Learning Latent Dynamics for Partially-Observed Chaotic Systemscited by 18
  12. 2021Beyond Predictions in Neural ODEs: Identification and Interventionscited by 15
10 more
  1. 2024Learning the Hodgkin-Huxley Model with Operator Learning Techniquescited by 10
  2. 2024Learning Physics Informed Neural ODEs With Partial Measurementscited by 9
  3. 2020Uncovering differential equations from data with hidden variablescited by 8
  4. 2024Learning Governing Equations of Unobserved States in Dynamical Systemscited by 7
  5. 2023Towards an Hybrid Hodgkin-Huxley Action Potential Generation Modelcited by 2
  6. 2026Geometric regularization of autoencoders via observed stochastic dynamicscited by 2
  7. 2015Parameter identifiability and identifiable combinations in generalized Hodgkin-Huxley modelscited by 1
  8. 2020Discovery of the Hidden State in Ionic Models Using a Domain-Specific Recurrent Neural Networkcited by 1
  9. 2025Correcting model error bias in estimations of neuronal dynamics from time series observationscited by 1
  10. 2026Learning Hybrid Biophysical Neuron Models with Neural ODEscited by 1

Built on by

Not cited yet.