Machine Learning with vvvv
How to make vvvv perceive the world by learning from data.
The present world is occupied more and more with devices and gadgets that claim to be "intelligent". They are able to recognize complex patterns or objects (e.g. face recognition in images), processes (e.g. hand gestures from tracking coordinates) and classes (e.g. music-genre recognition from audio data) and many more in a way that was until recently unique to the human brain.
Many of the methods behind these developments in general tagged as „machine learning“, a research field that made tremendous progress in the last years due to improvements in algorithms and a massive increase in available computing power. Using these techniques computers are able to interpret data on a much higher level than just raw values.
This workshops aims on introducing basic concepts of machine learning and pattern recognition and how it can be applied in VVVV using the Accord.net-framework. We will introduce the concepts of data preprocessing, feature extraction and selection, model training and post processing to gain knowledge of the higher concepts behind the raw data.
Since the area of machine learning itself is huge we will focus on a few practical examples of how they can be applied for applications like
- gesture recognition of touch or tracking data
- classification of various data (image, sound) into different categories
- and artistic use of these methods
- Basic Tutorials
- DX9 Rendering
- Basic knowledge of dynamic plugin programming
- We will try to avoid the math behind the concepts as much as possible. Yet some familarity with concepts of probability theory and statistics might be helpful.
- Windows laptop with latest vvvv version and addonpack installed