Limited Training Data Robust Speech Recognition Using Kernel-Based Acoustic Models | IEEE Conference Publication | IEEE Xplore