The Duke is well known in the Java universe. You might even have seen him flash by, Of course you would love to have a picture of the Duke, but he is rather shy. Even before the photographer can say ‘SMILE’, the Duke is already gone. We would like to help you to change this.
Within this HOL you will be using DL4J (Deep Learning 4 Java) to train a neural network into recognising the Duke. By using Transfer Learning, we will make sure that you can train your network within minutes. You will get a first experience in building neural networks, how you can train them and how you can improve your network.
To help you test your setup, all participants will be given a custom made a stuffed Duke figure. With the help of this figure you can make sure that you will be able to get a picture of the Duke the next time he flashes by. Off course you can take the Duke home to keep improving your software.
After this workshop you will have learned how to use Java in combination with DL4J to build and train a deep learning network. You will be able to create your own trainings data and retrain the network to recognize anything you want.
For this workshop you do not need any experience with Deep Learning. Basic programming knowledge should suffice.

Requirements:
• Java 1.8
• Maven
• Laptop with webcam