Artificial Intelligence as a research topic is a monopoly of researchers in Computer Science; and Artificial Intelligence and its various applications are a ripe field of possibilities, just not immediately accessible to someone not very familiar to Computer Science.
In this hands on workshop, we explore just some of the possibilities when we bring Artists and Artificial Intelligence together.
We begin the workshop by exploring possible use cases where AI can help manipulate the visual medium to provide a novel way of visual expression to the artist. More specifically we will cover Artistic Style Transfer, and AI augmented painting.
In the second quarter of the workshop, we will cover character level language models which can be used to generate "similar" text, given a corpus. In this case, we will train a model to help us write poetry.
In the third quarter of the workshop, we will extend the ideas from the language modelling exercise to music, with the goal of training a model which can compose music. We will train our model on a corpus of music written by Bach, and ideally the trained model should be able to generate more Bach-like music for us ;)
In the final quarter of the workshop, we will collaboratively review the results from the previous three exercises (at least the ones which are ready), and then finally end by brainstorming more creative ways to use Artificial Intelligence for Artists; and hopefully connect you with your collaborators for your next AI powered Art Installation.
Note that, for all the three use cases, we will provide starter code to help you get started, and also (hopefully) access to one computing node with a GPU.
In the best possible case, 1 painting that they collaboratively created with an AI, 1 poem that their baby Neural Network wrote for them; and 1 brand new Bach-like composition that their pet AI composed for them.
And in the worst case, a birds eye view of how some classic sub problems in DeepLearning work; and how they can help artists explore new mediums of expression.
Beginner level
- participants are not expected to know anything about DeepLearning or Artificial Intelligence in advance
- at least one team member should be proficient in "scripting" in python
- at least one team member should have super-human abilities when it comes to finding their way through Stackoverflow
- if you have a young child who loves to draw, and you have one of his drawings stuck on your fridge with a magnet, you **SHOULD** definitely bring a small picture of that drawing (or any of his/her drawings for that matter)