Aluminium profiles, silicone foil, neural network, computer, stepper motor, 3D-printed connection parts
The work System_0/1 is a space installation that tries to predict randomness. The installation constructs a coincidence and a learning process and investigates the interaction of these two systems. It examines the distinction between process and display and makes the process of learning visible within a system. The task of the system is to show how new information can be generated from different systems. For this purpose, a connection is established between an artificial neural network and a random generator, which becomes a model for the integration of randomness into a artificial neural network.
For this purpose, a connection is established between an artificial neural network (ANN) and a random generator, which becomes a model for the integration of randomness into ANN. The artistic work examines in a physical process how the influence of chance in iteration affects a self-learning system. In other words, how a contingency factor influences a 'non-trivial machine'.
The work ''System_0/1'' examines the relationship between an artificial neuronal system and chance. The system consists of two arranged systems (System_0 / System_1). Each of the two systems consists of a combination of motors, which can move an 'actuator' vertically arranged in a square. An elastic foil is stretched between the actuators. The 'actuator' measures the tension of the foil with a linear potentiometer, which varies between two springs.
System_0 functions as an input system for the artificial neural network. A self-referential coincidence arises out of the material.
System_1 consists of the process of adapting an artificial neural network, which processes the influences of System_0 into a self-learning simulation.
The system develops (learns) a generative output from the material. The stimulation of chance becomes the process of installation. A resulting surface consisting of four linear guides arranged in a square and a film stretched between them results in a random function which is determined by the residual stress of the foil.
The residual stress of the foil is measured by a linear potentiometer. In order to excite the system behavior, the information is taken from the residual stress of the System_0. The random values of System_0 flow into System_1 as entropy source. In iterations, the 'System_0/1' is checked for the function from the System_0 by means of the random value resulting from the material. The learning process of the ANN is represented by the adjustment and checking of the input information on system_1. This results in a system (System_0/1) in which the interaction between chance and learning is made visible. The task of the work is an investigation of complex interactions of entities in a kind of encapsulated space.