The Mechanical Neural Network(MNN) -- A physical implementation of a multilayer perceptron for education and hands-on experimentation

TL;DR

The paper introduces the Mechanical Neural Network (MNN), a physical, educational tool that models a multilayer perceptron using wooden levers and threads, enabling hands-on learning of neural network behavior and functions.

Contribution

It presents a novel physical implementation of an MLP for educational purposes, allowing direct manipulation of parameters and visualization of neural computations.

Findings

MNN can model real-valued functions.

MNN can implement logical operators including XOR.

Provides an interactive learning experience for neural networks.

Abstract

In this paper the Mechanical Neural Network(MNN) is introduced, a physical implementation of a multilayer perceptron(MLP) with ReLU activation functions, two input neurons, four hidden neurons and two output neurons. This physical model of a MLP is used in education to give a hands on experience and allow students to experience the effect of changing the parameters of the network on the output. Neurons are small wooden levers which are connected by threads. Students can adapt the weights between the neurons by moving the clamps connecting a neuron via a thread to the next. The MNN can model real valued functions and logical operators including XOR.