Meeting Banner
Abstract #2294

Data Downsampling for Time-Efficient and Robust Estimation of ICA Model Order Using Bootstrap Stability Analysis of Principal Components

Waqas Majeed1, Feng Wang2, 3, Robert M. Friedman4, Chaohui Tang2, 3, Malcolm J. Avison2, 3

1Department of Electrical Engineering, Lahore University of Management Sciences, School of Science and Engineering, Lahore, Punjab, Pakistan; 2Vanderbilt University Institute of Imaging Science (VUIIS), Vanderbilt University, Nashville, TN, United States; 3Department of Radiology, Vanderbilt University, Nashville, TN, United States; 4Department of Psychology, Vanderbilt University, Nashville, TN, United States

Bootstrap stability analysis (BSA) has previously been shown to be a more robust and accurate method for ICA model order estimation, compared with traditional methods. However, large computational burden associated with BSA makes it impractical. In this work, we demonstrate that BSA can be used on the prefiltered and downsampled data without compromising the results, while reducing the processing time to a great degree.

Keywords

achievable achieved activation actuators agreed agreement aliasing almost although anesthetized approved arises audience blocks blood bootstrap bootstrapping brain cerebral coated commonly communities complete component components compromising computation computational computed concatenated concatenation consists contrast correct correlation cortex cost courses criterion cross dataset datasets delivered density described digits driven effective efficient efficiently electrical engineering established estimation even expected explored falling feasible filter filtered filtering filters finger focal fold fractional frequency functional generate gift gradient greater hours hypothesis imposes improve indentation independent indicate individual influences injected kernel larger long loss making management maps model modes monkey monkeys motion much negligible noise null particle pass planar preconditioning prepared previous primary principal prior processing protocol radiology ranging real reduced reduces reductions related replicates required requiring response resting runs samples sampling satisfying school science sciences series session sessions seven significantly since skin stability stage stimulation stimulus studied studies suggested table tang target task temporal vertical volume volumes