Sparse Signal Recovery with Orthogonal Matching Pursuit

Introduction

Orthogonal Matching Pursuit (OMP) is a method for recovering a sparse signal from a noisy measurement encoded with a dictionary. In this lab, we will use scikit-learn to implement OMP to recover a sparse signal from a noisy measurement.

VM Tips

After the VM startup is done, click the top left corner to switch to the Notebook tab to access Jupyter Notebook for practice.

Sometimes, you may need to wait a few seconds for Jupyter Notebook to finish loading. The validation of operations cannot be automated because of limitations in Jupyter Notebook.

If you face issues during learning, feel free to ask Labby. Provide feedback after the session, and we will promptly resolve the problem for you.

Skills Graph

Import Libraries

import matplotlib.pyplot as plt
import numpy as np
from sklearn.linear_model import OrthogonalMatchingPursuit
from sklearn.linear_model import OrthogonalMatchingPursuitCV
from sklearn.datasets import make_sparse_coded_signal

Generate the Data

n_components, n_features = 512, 100
n_nonzero_coefs = 17

## generate the data

## y = Xw
## |x|_0 = n_nonzero_coefs

y, X, w = make_sparse_coded_signal(
    n_samples=1,
    n_components=n_components,
    n_features=n_features,
    n_nonzero_coefs=n_nonzero_coefs,
    random_state=0,
)
X = X.T

(idx,) = w.nonzero()

## distort the clean signal
y_noisy = y + 0.05 * np.random.randn(len(y))

Plot the Sparse Signal

plt.figure(figsize=(7, 7))
plt.subplot(4, 1, 1)
plt.xlim(0, 512)
plt.title("Sparse signal")
plt.stem(idx, w[idx])

Plot the Noise-Free Reconstruction

omp = OrthogonalMatchingPursuit(n_nonzero_coefs=n_nonzero_coefs)
omp.fit(X, y)
coef = omp.coef_
(idx_r,) = coef.nonzero()
plt.subplot(4, 1, 2)
plt.xlim(0, 512)
plt.title("Recovered signal from noise-free measurements")
plt.stem(idx_r, coef[idx_r])

Plot the Noisy Reconstruction

omp.fit(X, y_noisy)
coef = omp.coef_
(idx_r,) = coef.nonzero()
plt.subplot(4, 1, 3)
plt.xlim(0, 512)
plt.title("Recovered signal from noisy measurements")
plt.stem(idx_r, coef[idx_r])

Plot the Noisy Reconstruction with Number of Non-zeros Set by CV

omp_cv = OrthogonalMatchingPursuitCV()
omp_cv.fit(X, y_noisy)
coef = omp_cv.coef_
(idx_r,) = coef.nonzero()
plt.subplot(4, 1, 4)
plt.xlim(0, 512)
plt.title("Recovered signal from noisy measurements with CV")
plt.stem(idx_r, coef[idx_r])

plt.subplots_adjust(0.06, 0.04, 0.94, 0.90, 0.20, 0.38)
plt.suptitle("Sparse signal recovery with Orthogonal Matching Pursuit", fontsize=16)
plt.show()

Summary

In this lab, we have learned how to use Orthogonal Matching Pursuit (OMP) to recover a sparse signal from a noisy measurement encoded with a dictionary. We have used scikit-learn to implement OMP and generate a sparse coded signal. We have also plotted the sparse signal, the noise-free reconstruction, the noisy reconstruction, and the noisy reconstruction with the number of non-zeros set by CV.