AlexNet is considered as a standard architecture for a Convolutional Neural Network that is able to learn from big data. It is a relatively large CNN model with five convolution layers and three fully connected layers. The model is very popular and that is why it is often also available as pre-trained model like within MatConvNet for example. It is a pre-trained network trained on the ImageNet ILSVRC dataset that can be used as a starting point for many other learning problems.
We recommend to check out the following video about the subject: