🦜 Bird Classification using Perceptron (SLP & MLP) & ADALINE

A neural network-based bird species classification project implementing Perceptron (SLP), ADALINE, and Multi-Layer Perceptron (MLP) algorithms. The project includes data analysis, training & testing, implementing Mean Squared Error (MSE), activation functions, and a GUI for visualization.

---

🚀 Features

• ✅ Implementation of Single Layer Perceptron (SLP)
• ✅ Implementation of Multi Layer Perceptron (MLP)
• ✅ Implementation of ADALINE with MSE loss
• ✅ Custom training & testing (30/20 split per class)
• ✅ Data analysis (feature distribution, correlation, and visualizations)
• ✅ Graphical Interface (GUI) for user-friendly interaction
• ✅ Implementation of activation functions for model training

---

📦 Imports & Dependencies

This project requires the following Python libraries:

• numpy - For numerical operations
• pandas - For data manipulation and analysis
• matplotlib - For data visualization
• seaborn - For enhanced visualizations

You can install them using:

pip install numpy pandas matplotlib seaborn

---

📁 Project Structure

/bird-classification
│── main.py          # Main entry point to train and test models
│── slp.py           # Single Layer Perceptron Learning Algorithm (SLP)
│── mlp.py           # Multi Layer Perceptron Learning Algorithm (MLP)
│── adaline.py       # ADALINE Learning Algorithm (MSE)
│── utils.py         # Common functions (data loading, splitting, visualization)
│── gui.py           # GUI for visualization & interaction
│── data_analysis.py # Exploratory Data Analysis (EDA) & graphs
│── birds_data.csv   # Dataset (bird features & classes)
│── README.md        # Project documentation

📊 Data Analysis & Visualization

• 📈 Feature Distribution Histograms: Visualizes feature distribution and frequency.
• 📊 Class Distribution Plots: Shows class balance in dataset.
• 🔥 Correlation Matrix Heatmap: Displays feature pair correlations.
• 🧩 Decision Boundary Visualization: Visualizes model decision boundaries.
• 🥧 Gender Distribution Pie Chart: Shows gender proportions in dataset.
• 📊 Feature Distributions: Displays feature value distribution.
• 🔗 Scatter Plot Matrix: Shows pairwise feature relationships.
• 📦 Box Plots: Displays distribution, spread, and outliers.
• 🔍 PCA (Principal Component Analysis): Reduces dimensions, preserves variance.
• 🔄 Pair Plot: Grid of feature pair scatter plots.

---

🧠 Algorithms Implemented

1️⃣ Single-Layer Perceptron (SLP)

• Uses signum activation function
• Trained with perceptron learning rule
• Outputs binary class labels

2️⃣ Adaptive Linear Neuron (Adaline)

• Uses linear activation function during training
• Uses signum activation function during testing
• Optimized using Mean Squared Error (MSE)
• Outputs continuous values before thresholding

3️⃣ Multi-Layer Perceptron (MLP)

• Uses Sigmoid or tanh activation function during training
• Dynamic Number of Hidden Layers
• Outputs continuous values before thresholding

📊 Results & Performance

Model	Test Accuracy (%)
Perceptron	up to 100%
ADALINE	up to 100%
MLP	up to 100%

Best Features for ADALINE

• Best Combination: Body mass and Beak Depth for categories a and B.

Best Features for Perceptron (SLP)

• Best Combination: Beak depth and beak length for categories A and C are perfectly separable.