🔥 PyTorch Multi-Class Classification (Blobs Dataset)

This project demonstrates how to build, train, and evaluate a multi-class classification model using PyTorch from scratch. It walks through the full machine learning workflow—from data generation to model visualization.

---

📌 Project Overview

In this project, we generate a synthetic dataset using make_blobs and train a neural network to classify data points into 4 distinct classes based on their features.

The model learns to separate clusters in a 2D space and visualize decision boundaries.

---

⚙️ Key Features

• 📊 Synthetic dataset generation using sklearn.make_blobs
• 🔢 Data preprocessing and conversion to PyTorch tensors
• ✂️ Train-test split for proper evaluation
• 🧠 Custom neural network built with nn.Module
• 🔁 Training and testing loops from scratch
• 📉 Loss calculation using CrossEntropyLoss
• 🎯 Accuracy evaluation
• 📈 Prediction probabilities using softmax
• 🌍 Decision boundary visualization

---

🧠 Model Architecture

The model is a simple feedforward neural network with non-linear activation:

Input (2 features)
   ↓
Linear Layer (2 → 8)
   ↓
ReLU
   ↓
Linear Layer (8 → 8)
   ↓
ReLU
   ↓
Output Layer (8 → 4 classes)

---

🎯 Objective

The goal is to correctly classify each data point into one of the 4 classes based on its position in the feature space.

---

🚀 Learning Highlights

This project helps you understand:

• Difference between logits, probabilities, and predictions
• Why CrossEntropyLoss is used for multi-class classification
• How softmax + argmax converts model outputs into class labels
• Importance of non-linearity (ReLU) in neural networks
• How to structure a complete PyTorch training pipeline

---

📊 Results

• The model improves over epochs using gradient descent
• Decision boundary visualization shows how the model learns to separate classes
• Achieves strong classification performance on both training and test data

---

🖼️ Visualization

The project includes:

• Scatter plots of the dataset
• Model decision boundaries for both training and testing data

---

🧰 Tools & Libraries

• PyTorch
• Scikit-learn
• Matplotlib

---

💡 Why This Project Matters

This is a core deep learning project that builds a strong foundation for:

• Multi-class classification problems
• Neural network design
• Model evaluation and debugging
• Transitioning to real-world datasets

---

🔥 Next Steps

• Experiment with deeper networks
• Try different optimizers (Adam, RMSprop)
• Apply the model to real datasets
• Improve accuracy with hyperparameter tuning

---

💡 This project is a great stepping stone toward mastering deep learning and building real-world AI systems.