Inferring Latent Dynamics Underlying Neural Population Activity via Neural Differential Equations
Authors: Timothy D. Kim, Thomas Z. Luo, Jonathan W. Pillow, Carlos D. Brody
ICML 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We apply the PLNDE framework to a variety of synthetic datasets, and show that it accurately infers the phase portraits and fixed points of nonlinear systems augmented to produce spike train data... Our model significantly outperforms existing methods at inferring single-trial neural firing rates and the corresponding latent trajectories that generated them... and 4. Experiments We fit PLNDE (ours), PLDS1 (Section 2.1) and LFADS2 (Section 2.2) to synthetic datasets with three different nonlinear dynamical systems. We generated each latent trajectory (i.e., a trial) from a different initial value of the dynamical systems. We split the data into training and test, where test trajectories were generated from initial values not seen in training. |
| Researcher Affiliation | Academia | Timothy Doyeon Kim 1 Thomas Zhihao Luo 1 Jonathan W. Pillow 1 2 Carlos D. Brody 1 3 1Princeton Neuroscience Institute, Princeton, New Jersey 2Department of Psychology, Princeton University, Princeton, New Jersey 3Howard Hughs Medical Institute, Princeton University, Princeton, New Jersey. Correspondence to: Timothy Doyeon Kim <tdkim@princeton.edu>. |
| Pseudocode | No | The paper describes methods and processes but does not include any clearly labeled 'Pseudocode' or 'Algorithm' blocks or figures. |
| Open Source Code | No | The paper provides links to the code for comparison models (PLDS and LFADS) in footnotes 1 and 2, but does not provide a link or explicit statement about the availability of the source code for their own model, PLNDE. |
| Open Datasets | No | The paper uses 'synthetic datasets' which were generated by the authors, and for real data, it refers to 'multi-region neural population recordings from medial frontal cortex of rats performing an auditory decision-making task', citing Brunton et al. 2013 for the task. However, it does not provide a direct link, DOI, repository, or explicit statement of public availability for the specific dataset used in their experiments. |
| Dataset Splits | No | The paper states 'We split the data into training and test, where test trajectories were generated from initial values not seen in training.' It explicitly mentions training and test splits but does not mention a 'validation' split. |
| Hardware Specification | No | The paper does not specify any particular hardware components such as GPU or CPU models, or detailed specifications of the computing environment used for the experiments. |
| Software Dependencies | No | The paper mentions optimization methods like ADAM and model components like GRU, and references other works for neural ODEs, but it does not provide specific version numbers for any software dependencies, libraries, or frameworks used in the experiments. |
| Experiment Setup | Yes | We initialized the parameters of PLNDE with Glorot normal and trained for 5000 iterations. |