Transfer Learning :

Transfer learning is commonly used in deep learning applications. You can take a pretrained network and use it as a starting point to learn a new task. Fine-tuning a network with transfer learning is usually much faster and easier than training a network from scratch with randomly initialized weights. You can quickly transfer learned features to a new task using a smaller number of training images.

Some of existing neural networks for image classification for

-> AlexNet

-> Vgg16

-> Vgg19

-> ResNet-50

-> ResNet-101

-> GoogleNet

-> Inception-v3

-> Inception-ResNet-v2

In this implementation, two major steps are present:

Step 1: The last three layers : "Fully-Connected-Layer", "SoftMax" and "Classification Predictions" are removed.

Step 2: New three layers : "Fully-Connected-Layer", "SoftMax" and "Classification Predictions" are added but based on number of classes in our dataset.

Step 3: Connect Original Network's "Pooling-Layer" to newly created layers in Step 2

The difference in the using different neural network implementation (as given above) is defining the neural network model and the identification of these layers and replacing them.

References:

https://www.mathworks.com/help/nnet/ug/pretrained-convolutional-neural-networks.html