Consider the new set of features that you obtained by multiplying the PCA output with your feature set.
computer science
Description
A
Consider the new set of features that you obtained by multiplying the PCA output with your feature set.
Divide that new feature set into two parts for each user: a) part 1: training and b)
part 2: test. Ideally keep 60% of the data for each user as training and the rest of
40% as test data. Use three types of machines: a) decision trees (fitctree in
MATLAB), b) support vector machines (fitcsvm in MATLAB), and c) neural
networks (use the neural network toolbox in MATLAB).
Train each machine with the training data and then use the test data to report
accuracy. Use the accuracy metrics of Precision, Recall, F1 score. Report each
metric for every user.
#b
For a given gesture, consider 60% of total users and use all their feature points of
each user as training. Follow the same labelling strategy as considered in previous
user dependent analysis. The rest users are testing. Do the same analysis as in
previous case and report the same metrics for each of the rest of the test users.
