In at the deep end: Machine Learning with Java

Come to this Techday to learn more about Deep Learning and try training a model yourself!

Deep learning is a machine learning approach in which multiple layers are combined to transform raw input data into concepts that humans can interpret. Each layer is used to incrementally extract more advanced information based on the features provided by the previous layer.

This Techday will start off with our own experiences with AI and some real-life cases using AI. You will also learn more about the basics of machine learning and deep learning. After that, we will dive into advanced image recognition with deep learning using Deeplearning4j (an open-source library that allows you to combine simple neural nets in order to create a deep neural net). We will talk about the structure of deep neural nets and show you how to use this library. Given the amount of time needed to properly training a network, the possibility of using transfer learning is an interesting feature.

During the hands-on session, you can try it for yourself and train an image recognizer for one of the pre-defined datasets, or for your own images. For example: determine if your image contains an apple or an orange.

Requirements for your own dataset (We cannot be sure that a good working recognizer can be trained when using your own dataset. But it is more challenging!):

  • Maximum of 3 categories (for example cats, dogs, and mice)
  • The images must be in RGB format
  • Minimum of 50 images per category

Sign up via Meetup!

14:45 - 15:00 Doors open
15:00 - 15:30 Presentation
15:30 - 17:30 Hands-on 
17:30 - 18:30 Food!
18:30 - 19:00 Show your results