User:Alexander Roidl/notes
visual culture visual generation images
machine learning new learning algorithms
what to learn from machine learning and its influence on visual culture
Currently we encounter those strange new images, that look like morphing or artworks that are being generated by machine learning algorithms, without a deeper understanding of the underlying technology.
So machine learning became a hyped tool also for artists to do something "interesting".
But besides creating meaningless images by just throwing new datasets at those new algorithms, what can be learned from them? Especially, what has lasting potential to influence visual culture?
From generating random art to interventions in algorithms.
WHY? why why?
machine learning influences visual culture drop shadow did as well - so what is the point
machine learning changes how we perceive the world.
> gain a new understanding / meaning for visual material through new technology > new aesthetic / critic in arts, through visual production with machine learning >
The probability of an image
Generating images with statistical math.
Machine learning comes down to statistical calculations. So why is it so different from other algorithms? It makes decisions > hard to understand
Changes aesthetics > influence on image culture changes how we see
what is the relevance of the work?
Find a concrete issue
generation of photo-realistic images, image detection, touching surveillance and photography.
Not part of the dataset
what kind of images? generative images
Generated photographic images in 40 years
how will photo realistic image generation influence our life?
- stock images will be generated instead of downloaded
- advertising will change due to unlimited access to images
- there will be new fields instead of photographer > image generator (trying to make the machine dream the image, with different parameters)
- visualization will be possible without 3D animation / drawing / modeling
- image generation can happen in real time, that means we can live visualize
- generated footage and real world footage will not be distinguishable
- real and fake will fade
- new ways of differentiating real and generated images will be needed to preserve images as a proof / evidence
- image generation will become a own genre like photography
- photography will mostly be used for personal memory
- photography may not be substituted, as generation will be limited to generic uses
Visual Form of machine learning that intervene in the medium itself
- Software arts that generates Art
Learning aesthetic of machine learning
- Can machine learning help proof the concept of aesthetic / does it challenge it
What is my interest and why is it interesting?
Algorithmic automatic systems that influence our behaviour
Networks and how people interact with technology. Hidden structures, networks that are local but can be opened up
Datasets that are invisible but can be made visible and can be traced
Visual material / Analog material
Neural networks, how are these networks different
Neural networks are implemented in our daily life.
Comment on the use of neural networks.
I want to investigate on the inner functions of neural networks and their implications on a very low level.
While most reseach on machine learning is done on either a technical level, that tries to push the bounries, or on a cultural level, asking for major impacts of artificial intelligence, often drifting towards distopian critique, the algorithms and software itself are rarely subject to further research.
- Resources of machine learning
(In terms of computing power and also in terms of data)
- Construction of neural networks
- Hype
- Loss of control
- Decisionmaking