Perceptron Learning Visualizer

A Python implementation of the perceptron learning algorithm for educational purposes. Demonstrates how a single perceptron learns to approximate binary logic functions step by step.

Screenshots

Features

• Interactive command-line interface
• Support for any logical function using Python syntax
• Detailed step-by-step explanation of the learning process
• Visualizations of weights, biases, and thresholds
• Option to use simple perceptron or full perceptron with bias and learning rate
• Non-interactive demo mode showcasing multiple common logic functions

Run

uv sync

# Interactive mode
python script.py

# Demo mode (AND, OR, NOT, NAND, XOR)
python demo.py

# CLI
python cli.py --function "a and b" --detail detailed

Sample Output

PERCEPTRON CONFIGURATION
+----------------+----------+
| Logic Function | a and b  |
| Learning Rate  | 1        |
| Threshold      | 1.50     |
| Initial Bias   | -0.1     |
+----------------+----------+

Final weights: [1.00, 1.00]
Final bias: -0.10
Accuracy: 100.0%

Limitations

Single-layer perceptrons can only learn linearly separable functions:

• Can learn: AND, OR, NOT, NAND, NOR
• Cannot learn: XOR and other non-linearly separable functions

XOR demonstrates this limitation — the tool shows why it fails and explains multi-layer networks as the solution.