Positive sparse signal denoising: What does a CNN learn?

Mathematics

• Denoising 93%
• Neural Networks 67%
• Adaptive Thresholding 21%
• Multivariate Splines 17%
• Minimum Mean Square Error 16%
• Interpretability 15%
• Thresholding 14%
• Error Estimator 13%
• Regression Model 9%
• Interpretation 8%
• Demonstrate 7%
• Performance 7%
• Approximation 6%
• Generalization 6%
• Operator 5%

Engineering & Materials Science

• Convolutional neural networks 65%
• Splines 11%
• Mean square error 11%