A Learning-Based Framework for Line-Spectra Super-resolution
This work addresses spectrum estimation for signal processing applications, but it is incremental as it adapts existing learning techniques to a specific domain.
The authors tackled the problem of estimating the spectrum of multisinusoidal signals from limited samples by proposing a learning-based approach using neural networks trained on simulated data, which performs competitively with classical methods in Gaussian noise and is effective with impulsive noise.
We propose a learning-based approach for estimating the spectrum of a multisinusoidal signal from a finite number of samples. A neural-network is trained to approximate the spectra of such signals on simulated data. The proposed methodology is very flexible: adapting to different signal and noise models only requires modifying the training data accordingly. Numerical experiments show that the approach performs competitively with classical methods designed for additive Gaussian noise at a range of noise levels, and is also effective in the presence of impulsive noise.