🔢 MNIST Digit Classifier

A clean TensorFlow/Keras dense neural network that achieves high accuracy classifying handwritten digits from the MNIST dataset — with full training visualization and single-image prediction.

---

📌 Project Overview

This project trains a fully connected (dense) neural network on the classic MNIST dataset of 70,000 handwritten digit images. It covers the full supervised learning workflow: data loading, normalization, model building, training with validation monitoring, evaluation, and prediction.

A great first real-world deep learning implementation after understanding raw gradient descent.

---

📊 Key Results

Metric	Value
Dataset	MNIST (60K train / 10K test)
Model Type	Dense Neural Network (MLP)
Input Shape	28×28 grayscale → flattened
Output Classes	10 (digits 0–9)
Optimizer	Adam
Loss Function	Categorical Crossentropy

---

🔍 What This Lab Demonstrates

Concept	Implementation
Dataset loading	tf.keras.datasets.mnist
Preprocessing	Pixel normalization (÷255), one-hot encoding
Model architecture	Flatten → Dense(128, ReLU) → Dense(10, Softmax)
Training	Validation split, accuracy tracking
Evaluation	Test set loss and accuracy
Visualization	Training & validation curve plots
Prediction	Single test-image inference

---

🏗️ Model Architecture

Flatten(28 × 28)
    ↓
Dense(128, ReLU)
    ↓
Dense(10, Softmax)

---

🛠️ Tech Stack

Tool	Purpose
Python 3.x	Core language
TensorFlow / Keras	Model building and training
NumPy	Data manipulation
Matplotlib	Training curve visualization

---

🚀 How to Run

1. Clone the repo

git clone https://github.com/sainathac/mnist-digit-classifier.git
cd mnist-digit-classifier

2. Install dependencies

pip install -r requirements.txt

3. Run the lab

python src/main.py

---

📚 Part of the Deep Learning Lab Series

This is Lab 2 of 6 in a TensorFlow/Keras deep learning learning series:

1. Neural Network From Scratch
2. MNIST Digit Classifier ← You are here
3. Fashion-MNIST Model Lifecycle
4. Cats vs Dogs CNN Classifier
5. VGG16 Transfer Learning Classifier
6. IMDB Sentiment RNN

---

👤 Author

Sainatha C AI Automation & RPA Engineer | Data Science Practitioner