This repository contains implementations of the Adaline (Adaptive Linear Neuron) and Madaline (Multiple Adaptive Linear Neuron) networks, showcasing foundational neural network concepts through binary classification tasks.
- Adaline network implementation for binary classification with visualization of decision boundaries.
- Madaline network implementation with multiple neurons for enhanced pattern distinction.
- Demonstrates the use of different activation functions and learning rates.
- Includes detailed data generation, training processes, and visualization of model training and classification results.
- Clone the repository.
- Install the required Python libraries specified in
requirements.txt.
- Adaline is a simple type of single-layer neural network with weights adjusted according to the difference between the actual and predicted outputs (delta rule).
- The code provides functions to initialize the network, perform forward propagation, and apply weight updates.
- Madaline extends the Adaline by introducing multiple layers of neurons, allowing for more complex decision boundary formation.
- The code includes the mechanism for Madaline's forward propagation, decision making, and weight updates based on the training data.
- To run the Adaline network: execute the Adaline training function with the desired dataset and hyperparameters.
- For the Madaline network: initialize and run the Madaline training function, specifying the network architecture and training parameters.
- Both implementations output graphs showing the loss over training epochs, decision boundaries, and the classification's confusion matrix.
- The effectiveness of each model is quantified through performance metrics like accuracy and the mean squared error.
Contributions, bug reports, and feature requests are welcome! Feel free to fork the repository, make your changes, and submit a pull request.
This project is licensed under the MIT License - see the LICENSE file for more details.
- The AI and neural network community for foundational theories and practices that inspire these implementations.
For more information and to view the source code, visit the GitHub repository.