Dynamic Histogram Measurement for Time-Series Brain Signal Classification
Pattern classification based on time-series signals has attracted considerable attention in various domains. The conventional feature extraction approach used to extract statistical parameters from raw data has two drawbacks. One is the problem of lost information when patterns are represented by statistical parameter vectors. The other drawback is that statistical parameter vectors treat raw signal data as distributions, which do not capture dynamic information. Previous studies have identified that matching histograms perform better than matching vectors of statistical parameters extracted from raw data. Numerous other studies have explored how dynamic information can be captured. The main contribution of this thesis is to combine two approaches and suggest a better model. A Dynamic Histogram Measurement (DHM) for signal data is proposed, and the superiority of the new approach is assessed both theoretically and experimentally on electroencephalography (EEG) applications. The secondary contributions of this study include: (i) a comprehensive survey of EEG signal features, (ii) channel evaluation for EEG sensor placement, and (iii) a new database collection. The secondary contributions support the DHM approach to solving the proposed classification problem
Computer science|Medical imaging|Neurosciences
Li, Sukun, "Dynamic Histogram Measurement for Time-Series Brain Signal Classification" (2019). ETD Collection for Pace University. AAI27542566.
Remote User: Click Here to Login (must have Pace University remote login ID and password. Once logged in, click on the View More link above)