Orthogonal Pursuite de correspondance (OMP) is a algorithme glouton used in the field of apprentissage automatique and traitement du signal for solving sparse approximation problems. It aims to find a représentation sparse of a signal or data vector in a given dictionary of basis functions. The algorithm is particularly useful when the number of available measurements is much smaller than the number of potential basis functions, making it a popular choice in various applications such as compressed sensing and sélection de caractéristiques.
The core idea of OMP is to iteratively select the best matching elements from the dictionary that contribute most significantly to the approximation of the target signal. The process begins with an initial guess (often a zero vector) and, at each iteration, identifies the dictionary element that correlates most with the current residual (the difference between the actual signal and its approximation). This selected element is then added to the approximation set.
Une fois qu'un élément est choisi, l'OMP met à jour l'approximation en projetant le signal sur le sous-espace engendré par les éléments sélectionnés. Le résidu est alors recalculé, et le processus se répète jusqu'à ce qu'un nombre spécifié d'éléments ait été sélectionné ou jusqu'à ce que le résidu soit inférieur à un certain seuil. Cela rend l'OMP efficace en termes de calcul, car il réduit la taille du problème à chaque étape.
OMP is particularly valuable in applications where interpretability and simplicity are crucial, as it results in a sparse representation that highlights the most significant features of the data. However, it is important to note that OMP can be sensitive to noise and may not always guarantee the solution optimale, particularly in highly correlated dictionaries.