Digit-Recognizer

This code provides a comprehensive overview of building, training, and evaluating a Multi-Layer Perceptron (MLP) model for image classification on the MNIST dataset using PyTorch. Below is a summary of the major steps in the code:

1. Data Preparation:

   • Downloads the MNIST dataset using the TensorFlow Keras library.
   • Splits the training set into a new training set and a validation set.
   • Displays information about the dataset, such as the number of samples, minimum and maximum intensity values, and image shapes.

2. Data Loading:

   • Converts the dataset into PyTorch tensors.
   • Creates DataLoader objects for batching the data.

3. Model Architecture:

   • Defines an MLP model with three fully connected layers.
   • Moves the model to the GPU if available.

4. Model Summary:

   • Prints the shape of the output tensor for each layer of the model.

5. Optimizer:

   • Defines an SGD optimizer with a specified learning rate for updating model parameters.

6. Training Loop:

   • Trains the MLP model on the training data for a specified number of epochs.
   • Computes and collects training and validation losses for each epoch.
   • Saves the model checkpoints at the end of each epoch.

7. Learning Curve:

   • Plots the learning curve showing training and validation losses for each epoch.

8. Model Evaluation:

   • Loads the weights of the best epoch based on validation loss.
   • Tests the model on the test set.
   • Plots the predicted probabilities for a sample image.
   • Computes and prints the overall accuracy of the model on the test set.
   • Generates and displays a confusion matrix to evaluate model performance.

9. Final Thoughts:

   • Provides insights into the model's performance and suggests potential improvements.

This code serves as a useful reference for building and training MLP models for image classification tasks using PyTorch and offers guidance on evaluating model performance.