À suivreGuide suivant
Réseaux de neurones graphiques
Fondamentaux
GUIDE des fondamentaux
Convolutional neural networks use learned local filters shared across spatial positions to form feature maps.
Écoutez ce guideNarré par l'IA · ~3 min
The design supports image modeling, while the network's complete architecture and training determine its task performance.
A convolution layer applies a learned kernel at several spatial positions. At each position, it multiplies weights with corresponding input values and combines the products, commonly adding a bias. Deep-learning libraries often call this convolution even when the kernel is applied as an unflipped cross-correlation. Reusing a kernel across positions is weight sharing: a comparable layer does not learn a separate full kernel for every location. Local connectivity and shared weights are architectural properties, not a guarantee that a complete classifier recognizes an object equally well at every position. Padding, stride, pooling, boundary effects, and later layers all influence the result. Pooling commonly reduces spatial resolution and may reduce sensitivity to small local position changes, without guaranteeing complete translation invariance. Trained image networks often develop hierarchical features, with earlier patterns combined into more complex representations later, though every filter need not have a unique human-readable label. AlexNet is associated with the influential 2012 ImageNet classification result. Layer geometry is also practical to calculate. With a 6-by-6 input, 3-by-3 kernel, stride one, no padding, and no dilation, the kernel fits at four positions along each axis, giving a 4-by-4 output grid. Channels add another dimension. In a conventional dense convolution with three input channels and no grouping, one filter has kernel weights spanning all three channels and combines their contributions into one output feature map.
Training a classifier on a defined image dataset and evaluating it on held-out images.
Comparing learned feature maps across stages of a vision network.
Using local filters in an industrial camera-inspection experiment.
Studying how input channels, kernel sizes, padding, and stride change a convolution layer.
Convolution, attention, and hybrid designs can be compared for particular vision tasks and resource constraints. Report architecture, image resolution, training data, and evaluation conditions rather than assume one family universally dominates. Layer geometry and channel choices affect memory and computation, while accuracy needs separate measurement. Understanding the local shared-filter design helps explain both useful properties and the limits of treating every CNN as automatically invariant or reliable.
Newsletter gratuite
Trois histoires d'IA vérifiées chaque matin de la semaine, écrites dans un anglais simple. Gratuit pour toujours, sans publicité.
Un e-mail chaque jour de la semaine. Désabonnez-vous en un clic. Nous ne vendons ni ne partageons jamais votre adresse.
Testez-vous
Commentaires instantanés sur chaque réponse et un certificat partageable avec une pièce d'identité vérifiable une fois que vous avez réussi un cours.
Soutenez l’enseignement gratuit de l’IA. AI Understanding est une organisation à but non lucratif 501(c)(3) : pas de publicité, pas de paywall, jamais. Faire un don
Convolutional neural networks use learned local filters shared across spatial positions to form feature maps. The design supports image modeling, while the network's complete architecture and training determine its task performance.
Un filtre est une petite grille de poids qui glisse sur l'image et s'active lorsqu'elle trouve le motif qu'elle a appris, comme un bord ou une texture.
The comparison concerns shared versus separate local kernels; it does not claim every CNN is smaller than every fully connected model.
Pooling commonly downsamples; any added position tolerance is limited and depends on the design.
Les premières couches détectent les caractéristiques de bas niveau telles que les bords et les couleurs ; les couches plus profondes les combinent en concepts de niveau supérieur tels que des faces ou des parties d'objet.
The 2012 paper by Krizhevsky, Sutskever, and Hinton reports the network commonly known as AlexNet.
Continuez à apprendre
Plus de guides sélectionnés pour ce sujet
À suivreGuide suivant
Réseaux de neurones graphiques
Fondamentaux