while going through your paper " Pitch-density-based features and an SVM binary tree approach for multi-class audio classification in broadcast news ", I loved the approach; however, I am working on vibration signals for fault detection and want to implement the greedy backward elimination for feature selection. However, it appears its for regression-based algorithms but seeing your paper, I became more confident to use it for my project. could you help me with the source code?

Preferably, the code for the paper so I can understand better. I am an MSc student in Korea and would love to learn your approach.

Thanks in anticipation