{"id":477783,"date":"2023-08-09T09:20:08","date_gmt":"2023-08-09T09:20:08","guid":{"rendered":""},"modified":"2023-09-05T11:15:24","modified_gmt":"2023-09-05T11:15:24","slug":"k-nn-k-nearest-neighbours","status":"publish","type":"wiki","link":"https:\/\/oneproxy.pro\/fr\/wiki\/k-nn-k-nearest-neighbours\/","title":{"rendered":"k-NN (k-Voisins les plus proches)"},"content":{"rendered":"<p>Br\u00e8ves informations sur k-NN (k-Nearest Neighbours)<\/p>\n<p>k-Nearest Neighbours (k-NN) est un algorithme d&#039;apprentissage simple, non param\u00e9trique et paresseux utilis\u00e9 pour la classification et la r\u00e9gression. Dans les probl\u00e8mes de classification, k-NN attribue une \u00e9tiquette de classe bas\u00e9e sur la majorit\u00e9 des \u00e9tiquettes de classe parmi les \u00ab k \u00bb voisins les plus proches de l&#039;objet. Pour la r\u00e9gression, il attribue une valeur bas\u00e9e sur la moyenne ou la m\u00e9diane des valeurs de ses \u00ab k \u00bb voisins les plus proches.<\/p>\n<h2>L&#039;histoire de l&#039;origine de k-NN (k-Nearest Neighbours) et sa premi\u00e8re mention<\/h2>\n<p>L&#039;algorithme k-NN trouve ses racines dans la litt\u00e9rature sur la reconnaissance statistique de formes. Le concept a \u00e9t\u00e9 introduit par Evelyn Fix et Joseph Hodges en 1951, marquant le d\u00e9but de la technique. Depuis lors, il a \u00e9t\u00e9 largement utilis\u00e9 dans diff\u00e9rents domaines en raison de sa simplicit\u00e9 et de son efficacit\u00e9.<\/p>\n<h2>Informations d\u00e9taill\u00e9es sur k-NN (k-Nearest Neighbours). \u00c9largir le sujet k-NN (k-Nearest Neighbours)<\/h2>\n<p>k-NN fonctionne en identifiant les \u00ab k \u00bb exemples de formation les plus proches d&#039;une entr\u00e9e donn\u00e9e et en faisant des pr\u00e9dictions bas\u00e9es sur la r\u00e8gle de la majorit\u00e9 ou sur la moyenne. Les mesures de distance telles que la distance euclidienne, la distance de Manhattan ou la distance de Minkowski sont souvent utilis\u00e9es pour mesurer la similarit\u00e9. Les composants cl\u00e9s de k-NN sont\u00a0:<\/p>\n<ul>\n<li>Choix de &#039;k&#039; (nombre de voisins \u00e0 consid\u00e9rer)<\/li>\n<li>M\u00e9trique de distance (par exemple, euclidienne, Manhattan)<\/li>\n<li>R\u00e8gle de d\u00e9cision (par exemple, vote majoritaire, vote pond\u00e9r\u00e9)<\/li>\n<\/ul>\n<h2>La structure interne du k-NN (k-Nearest Neighbours). Comment fonctionne le k-NN (k-Nearest Neighbours)<\/h2>\n<p>Le fonctionnement de k-NN peut \u00eatre d\u00e9compos\u00e9 en les \u00e9tapes suivantes\u00a0:<\/p>\n<ol>\n<li><strong>Choisissez le chiffre &#039;k&#039;<\/strong> \u2013 S\u00e9lectionnez le nombre de voisins \u00e0 consid\u00e9rer.<\/li>\n<li><strong>S\u00e9lectionnez une mesure de distance<\/strong> \u2013 D\u00e9terminer comment mesurer la \u00ab proximit\u00e9 \u00bb des instances.<\/li>\n<li><strong>Trouver les k voisins les plus proches<\/strong> \u2013 Identifiez les \u00ab\u00a0k\u00a0\u00bb \u00e9chantillons de formation les plus proches de la nouvelle instance.<\/li>\n<li><strong>Faire une pr\u00e9diction<\/strong> \u2013 Pour la classification, utilisez le vote majoritaire. Pour la r\u00e9gression, calculez la moyenne ou la m\u00e9diane.<\/li>\n<\/ol>\n<h2>Analyse des principales caract\u00e9ristiques de k-NN (k-Nearest Neighbours)<\/h2>\n<ul>\n<li><strong>Simplicit\u00e9<\/strong>: Facile \u00e0 mettre en \u0153uvre et \u00e0 comprendre.<\/li>\n<li><strong>La flexibilit\u00e9<\/strong>: Fonctionne avec diverses mesures de distance et adaptable \u00e0 diff\u00e9rents types de donn\u00e9es.<\/li>\n<li><strong>Pas de phase de formation<\/strong>: Utilise directement les donn\u00e9es d&#039;entra\u00eenement pendant la phase de pr\u00e9diction.<\/li>\n<li><strong>Sensible aux donn\u00e9es bruyantes<\/strong>: Les valeurs aberrantes et le bruit peuvent affecter les performances.<\/li>\n<li><strong>Intensif en calcul<\/strong>\u00a0:\u00a0n\u00e9cessite le calcul des distances par rapport \u00e0 tous les \u00e9chantillons de l&#039;ensemble de donn\u00e9es d&#039;entra\u00eenement.<\/li>\n<\/ul>\n<h2>Types de k-NN (k-voisins les plus proches)<\/h2>\n<p>Il existe diff\u00e9rentes variantes de k-NN, telles que\u00a0:<\/p>\n<table>\n<thead>\n<tr>\n<th>Taper<\/th>\n<th>Description<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Norme k-NN<\/td>\n<td>Utilise un poids uniforme pour tous les voisins.<\/td>\n<\/tr>\n<tr>\n<td>k-NN pond\u00e9r\u00e9<\/td>\n<td>Donne plus de poids aux voisins les plus proches, g\u00e9n\u00e9ralement en fonction de l&#039;inverse de la distance.<\/td>\n<\/tr>\n<tr>\n<td>k-NN adaptatif<\/td>\n<td>Ajuste \u00abk\u00bb dynamiquement en fonction de la structure locale de l&#039;espace d&#039;entr\u00e9e.<\/td>\n<\/tr>\n<tr>\n<td>k-NN pond\u00e9r\u00e9 localement<\/td>\n<td>Combine \u00e0 la fois le \u00ab\u00a0k\u00a0\u00bb adaptatif et la pond\u00e9ration de la distance.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h2>Fa\u00e7ons d&#039;utiliser k-NN (k-Nearest Neighbours), probl\u00e8mes et leurs solutions li\u00e9es \u00e0 l&#039;utilisation<\/h2>\n<ul>\n<li><strong>Usage<\/strong>: Classification, R\u00e9gression, Syst\u00e8mes de recommandation, Reconnaissance d&#039;images.<\/li>\n<li><strong>Probl\u00e8mes<\/strong>: Co\u00fbt de calcul \u00e9lev\u00e9, Sensible aux fonctionnalit\u00e9s non pertinentes, Probl\u00e8mes d&#039;\u00e9volutivit\u00e9.<\/li>\n<li><strong>Solutions<\/strong>: S\u00e9lection des fonctionnalit\u00e9s, pond\u00e9ration de la distance, utilisation de structures de donn\u00e9es efficaces telles que KD-Trees.<\/li>\n<\/ul>\n<h2>Principales caract\u00e9ristiques et autres comparaisons avec des termes similaires<\/h2>\n<table>\n<thead>\n<tr>\n<th>Attribut<\/th>\n<th>k-NN<\/th>\n<th>Arbres de d\u00e9cision<\/th>\n<th>SVM<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Type de mod\u00e8le<\/td>\n<td>Apprentissage paresseux<\/td>\n<td>Apprentissage avide<\/td>\n<td>Apprentissage avide<\/td>\n<\/tr>\n<tr>\n<td>Complexit\u00e9 de la formation<\/td>\n<td>Faible<\/td>\n<td>Moyen<\/td>\n<td>Haut<\/td>\n<\/tr>\n<tr>\n<td>Complexit\u00e9 des pr\u00e9visions<\/td>\n<td>Haut<\/td>\n<td>Faible<\/td>\n<td>Moyen<\/td>\n<\/tr>\n<tr>\n<td>Sensibilit\u00e9 au bruit<\/td>\n<td>Haut<\/td>\n<td>Moyen<\/td>\n<td>Faible<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h2>Perspectives et technologies du futur li\u00e9es aux k-NN (k-Nearest Neighbours)<\/h2>\n<p>Les avanc\u00e9es futures pourraient se concentrer sur l\u2019optimisation de k-NN pour le Big Data, l\u2019int\u00e9gration de mod\u00e8les d\u2019apprentissage profond, l\u2019am\u00e9lioration de la robustesse au bruit et l\u2019automatisation de la s\u00e9lection des hyperparam\u00e8tres.<\/p>\n<h2>Comment les serveurs proxy peuvent \u00eatre utilis\u00e9s ou associ\u00e9s \u00e0 k-NN (k-Nearest Neighbours)<\/h2>\n<p>Les serveurs proxy, tels que ceux fournis par OneProxy, peuvent jouer un r\u00f4le dans les applications k-NN impliquant le web scraping ou la collecte de donn\u00e9es. La collecte de donn\u00e9es via des proxys garantit l&#039;anonymat et peut fournir des ensembles de donn\u00e9es plus diversifi\u00e9s et impartiaux pour cr\u00e9er des mod\u00e8les k-NN robustes.<\/p>\n<h2>Liens connexes<\/h2>\n<ul>\n<li><a href=\"https:\/\/scikit-learn.org\/stable\/modules\/neighbors.html\" target=\"_new\" rel=\"noopener nofollow\">Documentation Scikit-learn k-NN<\/a><\/li>\n<li><a href=\"https:\/\/en.wikipedia.org\/wiki\/K-nearest_neighbors_algorithm\" target=\"_new\" rel=\"noopener nofollow\">Page Wikip\u00e9dia sur l&#039;algorithme des k-voisins les plus proches<\/a><\/li>\n<li><a href=\"https:\/\/oneproxy.pro\/fr\/\" target=\"_new\" rel=\"noopener\">OneProxy \u2013 Solutions de serveur proxy<\/a><\/li>\n<\/ul>","protected":false},"featured_media":468739,"menu_order":0,"template":"","meta":{"_acf_changed":false,"content-type":"","inline_featured_image":false,"footnotes":""},"class_list":["post-477783","wiki","type-wiki","status-publish","has-post-thumbnail","hentry"],"acf":{"faq_title":"Frequently Asked Questions about <mark>k-NN (k-Nearest Neighbours)<\/mark>","faq_items":[{"question":"What is the k-Nearest Neighbours (k-NN) algorithm?","answer":"<p>The k-Nearest Neighbours (k-NN) is a simple and non-parametric algorithm used for classification and regression. It works by identifying the 'k' closest training examples to a given input and making predictions based on majority rule or averaging.<\/p>"},{"question":"What was the origin of the k-NN algorithm?","answer":"<p>The k-NN algorithm was introduced by Evelyn Fix and Joseph Hodges in 1951, marking its inception in statistical pattern recognition literature.<\/p>"},{"question":"How does the k-NN algorithm work?","answer":"<p>The k-NN algorithm works by choosing a number 'k', selecting a distance metric, finding the k-nearest neighbors to the new instance, and making a prediction based on majority voting for classification or computing the mean or median for regression.<\/p>"},{"question":"What are the key features of the k-NN algorithm?","answer":"<p>Key features of k-NN include its simplicity, flexibility, lack of a training phase, sensitivity to noisy data, and computational intensity.<\/p>"},{"question":"What are the different types of k-NN?","answer":"<p>There are various types of k-NN, including Standard k-NN, Weighted k-NN, Adaptive k-NN, and Locally Weighted k-NN.<\/p>"},{"question":"How can k-NN be used, and what are the related problems and solutions?","answer":"<p>k-NN can be used for classification, regression, recommender systems, and image recognition. Common problems include high computation cost, sensitivity to irrelevant features, and scalability issues. Solutions may involve feature selection, distance weighting, and utilizing efficient data structures like KD-Trees.<\/p>"},{"question":"How does the k-NN algorithm compare with other similar terms?","answer":"<p>k-NN differs from other algorithms like Decision Trees and SVM in aspects such as model type, training complexity, prediction complexity, and sensitivity to noise.<\/p>"},{"question":"What are the future prospects of k-NN?","answer":"<p>Future advancements in k-NN may focus on optimizing for big data, integrating with deep learning models, enhancing robustness to noise, and automating hyperparameter selection.<\/p>"},{"question":"How are proxy servers like OneProxy associated with k-NN?","answer":"<p>Proxy servers like OneProxy can be used in k-NN applications for web scraping or data collection. Gathering data through proxies ensures anonymity and can provide more diverse and unbiased datasets for building robust k-NN models.<\/p>"}]},"_links":{"self":[{"href":"https:\/\/oneproxy.pro\/fr\/wp-json\/wp\/v2\/wiki\/477783","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/oneproxy.pro\/fr\/wp-json\/wp\/v2\/wiki"}],"about":[{"href":"https:\/\/oneproxy.pro\/fr\/wp-json\/wp\/v2\/types\/wiki"}],"version-history":[{"count":0,"href":"https:\/\/oneproxy.pro\/fr\/wp-json\/wp\/v2\/wiki\/477783\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/oneproxy.pro\/fr\/wp-json\/wp\/v2\/media\/468739"}],"wp:attachment":[{"href":"https:\/\/oneproxy.pro\/fr\/wp-json\/wp\/v2\/media?parent=477783"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}