Table 2.

Strengths and limitations of deep learning and traditional machine learning methods for heart sounds classification.

Approaches	Strengths	Limitations
Tradition machine learning	1. Easy to train. 2. Can effectively and quickly solve the objective function by convex optimization algorithm.	1. Has a complex data preprocess and the segmenting of heart sound signal is indispensable. 2. Has generalization and robustness issues.
Deep learning	1. Can effectively and automatically learn feature representations and the trained model is very good generally. 2. Good performance in classification.	1. The training process takes a long-time and is affected by limited datasets. 2. High requirements for hardware configuration.