Handwritten Digit Classification Neural Model

This project is a neural network-based classifier for recognizing handwritten digits from images. The model is trained using TensorFlow and Keras and achieves high accuracy on the test set.

The handwritten digits classifier is built using a neural network with the following architecture:

• Input Layer: 784 units (corresponding to 28x28 pixel images)
• Hidden Layer 1: 400 units with ReLU activation
• Dropout Layer 1: Dropout rate of 0.1
• Hidden Layer 2: 250 units with ReLU activation
• Dropout Layer 2: Dropout rate of 0.1
• Output Layer: 10 units with softmax activation (corresponding to 10 digit classes) The model is trained on a dataset of 7000 images (60000 for training and 10000 for testing).

Dataset

The dataset consists of 28x28 pixel grayscale images of handwritten digits. Each image is represented as a flattened array of 784 values. The labels are integers from 0 to 9.

Requirements

• numpy
• matplotlib
• scikit-learn
• tensorflow
• seaborn

You can install the required libraries using pip:

pip install numpy matplotlib scikit-learn tensorflow seaborn