Bayesian Bi-clustering of Neural Spiking Activity with Latent Structures
Authors: Ganchao Wei
ICLR 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Validating our proposed MCMC algorithm through simulations, we find the method can recover unknown parameters and true bi-clustering structures successfully. We then apply the proposed bi-clustering method to multi-regional neural recordings under different experiment settings, where we find that simultaneously considering latent trajectories and spatial-temporal clustering structures can provide us with a more accurate and interpretable result. |
| Researcher Affiliation | Academia | Ganchao Wei Department of Statistical Science Duke University Durham, NC 27708, USA ganchao.wei@duke.edu |
| Pseudocode | No | The paper describes the MCMC algorithm steps in numbered lists within the text, but it does not present them as structured pseudocode blocks or explicitly labeled algorithm figures. |
| Open Source Code | Yes | The Python implementation of the NB and Poisson bi-clustering model is available in https: //github.com/weigcdsb/bi_clustering and supplementary material. |
| Open Datasets | Yes | We then apply our bi-clustering method to Allen Institute Visual Coding Neuropixels dataset. The dataset contains neural spiking activity from multiple brain regions of an awake mouse, under different visual stimuli. See Siegle et al. (2021) for more detailed data description. |
| Dataset Splits | No | The paper describes validating its method on a simulated dataset and applying it to a real dataset (Allen Institute Visual Coding Neuropixels dataset), but it does not specify explicit train/validation/test splits for these datasets as typically done for supervised learning models. The problem is a clustering problem, not a predictive one, so these splits are not relevant in the usual sense. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware used to run the experiments, such as GPU or CPU models. |
| Software Dependencies | No | The paper states 'The Python implementation... is available', but it does not provide specific version numbers for Python or any other key software dependencies or libraries used. |
| Experiment Setup | Yes | In this simulation, we generate 3 clusters with 10 neurons in each cluster (N = 30 in total). The recording length is T = 500 and the dimension for x(j) 1:T are all p = 2. For each neuron, the individual baseline is generated by di N(0, 0.52), the factor loading is generated by ci N(0, I2) and dispersion are all ri = 10. For latent factors... generated from two discrete states... The bias term is b = 0 and noise covariance is Q = I9 10 2 for both states. We here run 10,000 iterations using p = 2 and m = 10, simply to illustrate the usage of proposed method on real data. Since these neurons come from three brain regions, we set the prior for number of subject cluster as k Geometric(0.415) , such that P(k 3) = 0.8. |