Use svm to train ijcnn1 and test ijcnn1.t. Due to its large size, you should select a subset of ijcnn1 first for parameter selection (using the tune() function). Then you use few parameter sets on the whole set.
The R svm is simply an interface to libsvm. We would like to compare the time difference between c++ binaries and the R code. To use libsvm directly, you can follow the following guide. You then compare the running time of the two and discuss why some difference exists.
After obtaining the accuracy using svm, you would like to compare it with that from earlier homeworks.