Acoustic Scene Classification and Visualization of Beehive Sounds Using Machine Learning Algorithms and Grad-CAM

Table 2

Performance comparison of different features using various machine learning algorithms.

Model	Features	Parameters	Accuracy (%)
SVM	Mel spectrogram	C = 0.01, gamma = 0.1	77.58
	MFCCs	C = 10, gamma = 0.01	85.63
	CQT	C = 0.1, gamma = 0.1	77.55
Random Forest	Mel spectrogram	n_estimators = 100	82.14
		max_depth = 6
		min_samples_leaf = 3
	MFCCs	n_estimators = 100	86.82
		max_depth = 8
		min_samples_leaf = 5
	CQT	n_estimators = 200	74.07
		max_depth = 12
		min_samples_leaf = 3
XGBoost	Mel spectrogram	n_estimators = 200	82.98
		learning_rate = 0.01
		max_depth = 4
	MFCCs	n_estimators = 800	87.36
		learning_rate = 0.01
		max_depth = 8
	CQT	n_estimators = 800	74.35
		learning_rate = 0.2
		max_depth = 20