DataScience_Project

The repository contains projects related to data science.

---

Neural Network for Image Recognition of Letters A, B, and C

This project demonstrates a simple neural network trained to recognize images of letters A, B, and C using Python and NumPy.

Dataset

The dataset consists of three images:

• A: Represented as a list a
• B: Represented as a list b
• C: Represented as a list c Each letter is encoded as a binary image in a 1-dimensional NumPy array format.

Labels

The labels for the images are:

• A: [1, 0, 0]
• B: [0, 1, 0]
• C: [0, 0, 1]

Neural Network Architecture

• Input Layer: 30 neurons (corresponding to 30 pixels in each image)
• Hidden Layer: 5 neurons, activated using the sigmoid function
• Output Layer: 3 neurons (one for each letter), activated using the sigmoid function

Training

The neural network is trained using backpropagation with gradient descent -

• Loss Function: Mean Squared Error (MSE)
• Activation Function: Sigmoid Training Parameters:
• Learning Rate (α): 0.1
• Epochs: 100

Results

After training for 100 epochs, the neural network achieved an accuracy of 95.67% on the training dataset.

Usage

To use the trained model for prediction:

Test the Neural Network

Making Predictions Using Reshaped Arrays

print("Testing the trained neural network:")
test_samples = [np.array(a).reshape(1,30), np.array(b).reshape(1,30), np.array(c).reshape(1,30)]
for sample in test_samples:
    predict(sample, w1, w2)

Predicting using individual samples from a dataset

predict(x[0], w1, w2)  # Image is of letter A
predict(x[1], w1, w2)  # Image is of letter B
predict(x[2], w1, w2)  # Image is of letter C

Visualizations

Accuracy over Epochs

Loss over Epochs

---