We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

stat.ML

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Statistics > Machine Learning

Title: Nonlinear Evolution via Spatially-Dependent Linear Dynamics for Electrophysiology and Calcium Data

Abstract: Latent variable models have been widely applied for the analysis of time series resulting from experimental neuroscience techniques. In these datasets, observations are relatively smooth and possibly nonlinear. We present Variational Inference for Nonlinear Dynamics (VIND), a variational inference framework that is able to uncover nonlinear, smooth latent dynamics from sequential data. The framework is a direct extension of PfLDS; including a structured approximate posterior describing spatially-dependent linear dynamics, as well as an algorithm that relies on the fixed-point iteration method to achieve convergence. We apply VIND to electrophysiology, single-cell voltage and widefield imaging datasets with state-of-the-art results in reconstruction error. In single-cell voltage data, VIND finds a 5D latent space, with variables akin to those of Hodgkin-Huxley-like models. VIND's learned dynamics are further quantified by predicting future neural activity. VIND excels in this task, in some cases substantially outperforming current methods.
Comments: 8 figs, Accepted at NBDT
Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Neurons and Cognition (q-bio.NC); Quantitative Methods (q-bio.QM)
Cite as: arXiv:1811.02459 [stat.ML]
  (or arXiv:1811.02459v3 [stat.ML] for this version)

Submission history

From: Daniel Hernandez Diaz [view email]
[v1] Tue, 6 Nov 2018 16:10:56 GMT (451kb,D)
[v2] Tue, 25 Jun 2019 16:18:42 GMT (600kb,D)
[v3] Tue, 16 Jun 2020 18:00:58 GMT (1415kb,D)

Link back to: arXiv, form interface, contact.