An Investigation to Improve Classifier Accuracy for Myo Collected Data

Report No. ARL-TN-0813
Authors: Michael H Lee, Andre V Harrison, and Robert P Winkler
Date/Pages: February 2017; 64 pages
Abstract: A naÏve Bayes classifier trained with 1,360 samples from 17 volunteers performs at an accuracy of 72.5% (based on 10-fold cross validation). This accuracy is based on using the entire data set. One approach to increasing the accuracy is by analyzing the data and removing irregular samples from the training set. As the quality of the training data increases, the accuracy of the classifier will increase. This report describes analysis of 2 features of the training data, observed unusual patterns, and how fine tuning the training set increased the accuracy by 7.2%.
Distribution: Approved for public release
  Download Report ( 1.587 MBytes )
If you are visually impaired or need a physical copy of this report, please visit and contact DTIC.

Last Update / Reviewed: February 1, 2017