Convolutional Neural Network (CNN)
Definition
A composite function obtained by appropriately combining a convolutional layer, pooling layer, and activation function is called a convolutional neural network.
Description
A function composed of a convolutional layer and an activation function is called a convolutional neural network. CNNs demonstrate excellent performance predominantly in image-related tasks. In the case of MLP, when values pass through each layer, they are sent to a fully connected layer, which can lead to an enormous number of parameters if the data is high-dimensional and the network deepens. Conversely, in convolutional layers, the number of parameters depends solely on the kernel size, irrespective of the input data’s size. This allows for a significant reduction in the number of parameters compared to linear layers.
Historically, CNNs were proposed by mimicking how the visual cortex operates in the brain.
Types
- LeNet (1998)
- AlexNet (2012)
- VGG (2014)
- GoogLeNet (2014)
- ResNet (2015)
- U-Net (2015)
- DenseNet (2017)