The Mechanical Neural Network (MNN)

Adjusting the weights of the MNN to learn the XOR problem. Here the first fully wooden version of the MNN.

The Mechanical Neural Network is a mechanical implementation of an artificial neural network. More precisely, it is a Multilayer-Perceptron (MLP) with ReLU activation functions. It can be trained to model real-valued functions, classification tasks, and logical functions such as AND, OR, and the exclusive OR (XOR). This means we can adjust the weights between the neurons of the network so that the network produces the correct output for a given input. For example, we can adjust the weights so that the network models the XOR function and outputs true if one of the two inputs is true and the other is false, or classify animals according to properties such as body size and habitat into elephant, dog, whale, and fish.

It is built from wooden levers that represent the neurons. These levers are connected by strings that correspond to the connections between the neurons of an MLP. The manual adjustment of the sliders that connect the strings to the levers allows us to adjust the weights of the network. Thus, the effect of weight adjustment can be intuitively observed. While the first version was still made of wood, the following versions are colorfully 3D-printed, making it easier to assign weights and neurons.

Would you like your own MNN for example for teaching or workshops? Are you interested in a workshop where I teach the basics of AI with the MNN? Contact me at mnn(at)schaffland.eu

The 3D-printed version of the MNN prepared for animal classification.

A Modular System for Teaching AI with the MNN

With the MNN, various aspects of artificial neural networks can be explained at different levels of detail. Through experiential learning, learners of all age groups can experience the basics of AI without any computer or programming knowledge.

The building blocks that enable the teaching of various aspects of AI with the MNN.

Further Information

More information can be found in the explanatory video (English) on YouTube:

Or in one of the following publications:

Promoting the digital transformation of STEM education with the mechanical neural network, a physical model for future-oriented and student-centered AI education: A. Schaffland, C. Müller, and J. Schöning, 2024, IEEE Global Engineering Education Conference (EDUCON), IEEE, DOI: 10.1109/EDUCON60312.2024.10578714
Mechanical Neural Network: Making AI comprehensible for everyone: A. Schaffland and J. Schöning, 2023, IEEE German Education Conference (GeCon), IEEE, DOI: 10.1109/gecon58119.2023.10295144

News

News about the MNN can be found here.

Images