Inferring sparse representations of continuous signals with continuous orthogonal matching pursuit

Authors: Karin C Knudson, Jacob Yates, Alexander Huk, Jonathan W Pillow

NeurIPS 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We test the resulting method, which we call Continuous Orthogonal Matching Pursuit (COMP), on simulated and neural data, where it shows gains over CBP in both speed and accuracy.
Researcher Affiliation Academia Karin C. Knudson Department of Mathematics The University of Texas at Austin kknudson@math.utexas.edu Jacob L. Yates Department of Neuroscience The University of Texas at Austin jlyates@utexas.edu Alexander C. Huk Center for Perceptual Systems Departments of Psychology & Neuroscience The University of Texas at Austin huk@utexas.edu Jonathan W. Pillow Princeton Neuroscience Institute and Department of Psychology Princeton University pillow@princeton.edu
Pseudocode No No pseudocode or clearly labeled algorithm block was found.
Open Source Code No The paper does not provide concrete access to source code for the methodology described.
Open Datasets No The paper uses simulated data, which is generated internally, and neural data collected by the authors without providing public access information.
Dataset Splits No The paper describes using simulated and neural data and evaluating performance directly on it, but does not specify explicit train/validation/test dataset splits in the conventional sense.
Hardware Specification No No specific hardware details (e.g., GPU/CPU models, processor types) used for running experiments were mentioned.
Software Dependencies No The paper mentions "the cvx package for MATLAB" but does not specify a version number in the main text directly in relation to the authors' implementation. Reference [7] mentions "version 2.0" for CVX, but this is not explicitly stated as a dependency with its version number for the authors' experimental setup.
Experiment Setup Yes Throughout, we use K = 3, since the polar basis requires 3 basis vectors per bin. We categorize hits, false positive and misses based on whether a time shift estimate is within a threshold of ϵ = 1 of the true value. ... we also impose a thresholding step after recovery with CBP, discarding any recovered waveforms with amplitude less than .3.