FROM DATA TO DECISION: A TUTORIAL ON ARTIFICIAL INTELLIGENCE

Abstract

Artificial intelligence (AI) and machine learning (ML) are widely used in modern science and technology. Among various classification algorithms, the Rosenblatt perceptron is one of the earliest and simplest trainable models, forming the foundation of modern neural networks. This work presents a concise introduction to the perceptron model, including its structure, signum activation function, and error-correction learning rule. A physics-motivated worked example drawn from Tesla coil wireless power transfer demonstrates successful training and convergence of the algorithm: the perceptron learns to predict whether a receiver LED glows ($+1$) or does not glow ($-1$) based solely on the transmitter voltage and the coil-to-LED distance. A Python implementation is also provided to illustrate the development of a basic AI/ML classifier from scratch. The results show that the perceptron is an effective and interpretable method for binary classification, and the learned weight vector acquires a direct physical meaning in terms of the electromagnetic coupling strength.