Convolutional neural network
In machine learning, a convolutional neural network (CNN, or ConvNet) is a class of deep, feed-forward artificial neural networks, most commonly applied to analyzing visual imagery.
Addition to the normal neural network lies in usage of convolution, where instead of 1-1 connection (N-N most likely) between neurons, we shrink the number of required internal variables with a convolution process. This process is described in my image processing toturial : http://education.pkodytek.com/image-processing/03_convolution/
In this exercise we will use pre trained convolutional neural network from google to classify images of different items.
Example of image classification with its confidence
Top 5 predictions
| Data | Value |
|---|---|
| Source | https://www.mathworks.com/help/nnet/examples/classify-image-using-googlenet.html |
| Reference | https://en.wikipedia.org/wiki/Convolutional_neural_network |
| Code | 13_CNN_solution_code.zip |