Latent Dynamic Factor Analysis of High-Dimensional Neural Recordings

Authors: Heejong Bong, Zongge Liu, Zhao Ren, Matthew Smith, Valerie Ventura, Robert E. Kass

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Simulations show that LDFA-H outperforms existing methods in the sense that it captures target factors even when within-region correlation due to noise dominates cross-region correlation. We applied our method to local field potential (LFP) recordings from 192 electrodes in Prefrontal Cortex (PFC) and visual area V4 during a memory-guided saccade task. The results capture time-varying lead-lag dependencies between PFC and V4, and display the associated spatial distribution of the signals.
Researcher Affiliation Academia Heejong Bong , Zongge Liu Carnegie Mellon University hbong, zonggel@andrew.cmu.edu Zhao Ren University of Pittsburgh zren@pitt.edu Matthew A. Smith, Valérie Ventura, Robert E. Kass Carnegie Mellon University msmith, vventura, kass@andrew.cmu.edu
Pseudocode No The paper refers to Appendix A for full formulations of the EM algorithm but does not contain any pseudocode or algorithm blocks in the main text.
Open Source Code Yes The code is provided at https://github.com/Heejong Bong/ldfa.
Open Datasets Yes We now report the analysis of LFP data in areas PFC and V4 of a monkey during a saccade task, provided by Khanna et al. (2020).
Dataset Splits Yes We determine the remaining parameters λcross and q by 5-fold cross-validation (CV).
Hardware Specification Yes a single iteration of E and M-steps on our cluster server (with 11 Intel(R) Xeon(R) CPU 2.90GHz processors)
Software Dependencies No No specific software dependencies with version numbers are listed in the paper.
Experiment Setup Yes We applied LDFA-H using hauto = hcross = 10, corresponding to 100 ms (at 100 Hz)... The other tuning parameters were determined by 5-fold CV over λcross {0.0002, 0.002, 0.02, 0.2} and q {5, 10, 15, 20, 25, 30}, yielding optimal values λcross = 0.02 and q = 10.