Project Neural Network

The video tutorial explains how to use Keras with TensorFlow to build a neural network for classifying handwritten digits. It covers the installation of necessary libraries, data preparation, and preprocessing. The tutorial demonstrates building a simple neural network using the Sequential Model API, training it, and evaluating its performance. It also discusses adding hidden layers to improve accuracy and analyzing predictions using a confusion matrix.

What is the primary advantage of using Keras over TensorFlow for neural network design?

• Keras allows for designing neural networks with fewer lines of code.

• Keras supports more data types than TensorFlow.

• Keras provides more control over neural network design.

• Keras is faster than TensorFlow.

Which dataset is used in the tutorial for classifying handwritten digits?

• ImageNet

• CIFAR-10

• MNIST

• Fashion-MNIST

Why do we divide pixel values by 255 during preprocessing?

• To reduce the size of the dataset.

• To convert images to grayscale.

• To increase the pixel values.

• To normalize the pixel values for better processing.

What is the purpose of the Sequential Model API in Keras?

• To allow parallel processing of layers.

• To visualize the neural network architecture.

• To build neural networks layer by layer in sequence.

• To automatically optimize the model.

Which activation function is used in the output layer of the simple neural network?

• Softmax

• ReLU

• Tanh

• Sigmoid

What is an epoch in the context of training a neural network?

• A single update of the model weights.

• A single pass through the entire training dataset.

• A single iteration over a batch of data.

• A single evaluation of the model on test data.

What is the purpose of evaluating the model on test data?

• To adjust the model's hyperparameters.

• To verify the model's ability to generalize to new data.

• To increase the model's training accuracy.

• To reduce the model's training time.

What is a potential downside of training a neural network for too many epochs?

• Increased training speed

• Decreased model complexity

• Overfitting

• Underfitting

What is the effect of adding a hidden layer to the neural network?

• It simplifies the model's architecture.

• It decreases the model's complexity.

• It reduces the number of trainable parameters.

• It increases the model's accuracy.

Which activation function is commonly used in hidden layers for image data?

• Linear

• Sigmoid

• ReLU

• Softmax