In our daily lives, we increasingly encounter systems that seemingly have the ability to “think” or “make decisions.” Voice assistants, image and text recognition, and automated vehicles are just a few examples. But how do these intelligent systems work? At their core is neural network technology, which was inspired by the human brain and implemented in software and hardware.
Explanation
Neural networks consist of numerous “artificial neurons” that are interconnected. These networks learn from examples. They “see,” “hear,” or “measure” data, recognize patterns, and make predictions or decisions based on them. In this way, a machine is not only enabled to work rigidly according to rules but also to react flexibly to new situations.
Exhibit:
What is happening here?
The exhibit playfully demonstrates how an artificial neural network works. Here, you can step into the role of a “trainer” and experience how machines learn to distinguish between objects—in this case, a rabbit and a frog.
How it works:
- At the top of the diagram are input sliders that can be used to adjust the animals’ characteristics, such as “long ears,” “fluffy tail,” or “nose.”
- The weight can be adjusted using the cloud slider. This determines how strongly a feature influences the result.
- The nodes in the middle are the “neurons.” They collect light (information) from the incoming connections, process it, and pass it on.
- Ultimately, the network displays the probability of whether the input characteristics belong to a rabbit or a frog.
What you learn:
- How information flows in an artificial brain.
- That machines do not actually know what a rabbit or a frog is, but only calculate probabilities based on data.
- Even small changes can completely alter the result, just as with human perception.
Goal of the exhibit:
The exhibit illustrates the fundamentals of artificial intelligence (AI) and machine learning.
It originates from the THI project “Neural Networks” as part of the “Mensch in Bewegung” research initiative and is intended to show that AI is not magic, but is based on mathematics, logic, and extensive training.
