Neural Synthesis
- Costs: 55€ (4h / 1day)
- Max number of participants: 24 people
- 50% fee via Eventbrite: (left part to be paid in cash)
- Contact:
the workshop is part of the program of the Exhibition #13: Cybernetic Choreographies
This workshop will introduce the technique of synthesizing images from the imagination of a deep convolutional neural network, better known as "deepdream" or neural synthesis and visualization.
Using the neural synthesis technique, we can reverse a neural network's normal operation to have it draw patterns and objects recalled from its memory by stimulating its individual neurons and layers. We will learn how to mix, distort, and combine these extremely varied, colorful, and diverse images into unique compositions.
We will also briefly show how to generate animated videos, a subject which will be covered in more detail in a future follow-up workshop TBA. No prior experience with machine learning or programming is required, as the code will be provided.
All participants will make several of their own creations with the tools provided before the end of the class. A gallery of some example images can be found on
about the workshop holder
Gene Kogan is an artist and a programmer who is interested in generative systems, computer science, and software for creativity and self-expression. He is a collaborator within numerous open-source software projects, and gives workshops and lectures on topics at the intersection of code and art. Gene initiated ml4a, a free book about machine learning for artists, activists, and citizen scientists, and regularly publishes video lectures, writings, and tutorials to facilitate a greater public understanding of the subject.