GMU:Art in the times of surveillance capitalism – Understanding machine learning video-classification

From Medien Wiki
Revision as of 09:43, 11 April 2023 by Koenig (talk | contribs) (Created page with "The topic of AI is discussed in the media in terms of artificial consciousness, while the actual machine-learning applications have long been an integral part of the success o...")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

The topic of AI is discussed in the media in terms of artificial consciousness, while the actual machine-learning applications have long been an integral part of the success of IT giants. To understand this technology, it is essential to understand the principles of modern network architecture and its data structure and transmission. The course will take a practical approach to this topic and then lead into an informed discussion that will go beyond the opinions of "gift book philosophers".

The course gives an introduction to machine learning and its programming in Python using Nvidia Jetson Nano Computers set up as a Cloud-Cluster, which we set up in the seminar.


Example for a subject of discussion: Sentiment Analysis

There is currently no scientific consensus on a definition of emotions. But for sure they cannot be measured like temperature or height. They heavily differ individually through complex cultural, political, and historical influences.

So keep in mind that, by using such technology without any layer of reflection and critique, you transport reductionistic behavioral ideas that are the reason for discrimination and the disadvantage of people that don't fall within the statistical norm of such machine learning systems.

Content:

Every student will be provided with a NvidiaJetson Nano Developer Kit, from the university (as a rental). https://developer.nvidia.com/embedded/jetson-nano-developer-kit

You get a basic understanding of:

- Linux operating system - Network structures - Video processing (OpenCV and ffmpeg)

- machine learning (basic models) - classification with neural networks

optional (advanced):

- docker containers

Though the general outline of the seminar is fixed, certain topics can be adapted to the demands of the projects and the wishes of the students.

Programming knowledge in Python is mandatory.

The aim of the course is to gain a critical understanding of machine learning and its application. The course focuses on the analysis of the classification of video streams and their classification.

Another central topic is cloud infrastructures and the so-called "edge computing" or "Internet of Things", which together with machine learning, form an almost all-encompassing set of tools for data collection that is beyond any (state) control. The course is therefore also suitable for those who are interested in a critical examination of "AI".

Dr. phil. Alexander König - www.media-art-theory.com