CNN(Convolutional Neural Network)

Check this great explanation out:

https://www.youtube.com/watch?v=KuXjwB4LzSA&t=550s

Convolutional neural networks imitate the structure of neurons that process images in the brain and use techniques to reduce neuron count, as well as maintaining positional relationships in the data.

Convolutional Neural Network

Convolutional Neural Networks (CNNs) are inspired by the structure of neurons in the human brain responsible for processing visual information. They employ techniques to reduce the number of neurons while preserving spatial relationships within the data.

In this project, we've implemented a CNN to work with the MNIST dataset, aiming to recognize handwritten digits and explore different approaches for improved performance.

Using 10 Epochs as a standard base, we are able to achieve 99.25% accurcy in recognizing MNIST images.

Data Preparation

First, we import and prepare the MNIST dataset for our CNN. We reshape the images, normalize the pixel values, and one-hot encode the labels to prepare the data for training.

Creating Your Network

We set up the parameters for our CNN, including the number of epochs for training. We use the Sequential model from Keras to define our network.

Adding Layers

The CNN consists of convolutional layers, pooling layers, dropout layers, and dense layers. Convolutional layers use filters to extract features from the input data. Pooling layers reduce complexity and improve translation invariance. Dropout layers prevent overfitting. Dense layers make the final classification decision.

Compiling the Network

We compile the network using categorical cross-entropy as the loss function, the RMSProp optimizer, and accuracy as the evaluation metric.

Training

We train the model on the MNIST training data, using batch processing and shuffling for better convergence. We monitor the performance on the validation data during training.

Evaluating and Returning the Model

After training, we evaluate the model's performance on the test dataset and print the test accuracy.

Saving your Model

Finally, we save the trained model to a file named 'cnn_model.h5' for future use.

Follow the cnn.ipynb file to train your own CNN.

Credits https://www.tensorflow.org/tutorials/images/cnn https://www.youtube.com/watch?v=KuXjwB4LzSA