Building the Neural Network — Flashcards

Answer

Keras is a high-level neural network API in Python that simplifies constructing and training neural networks. It is used for tasks involving image and video data, text and audio data, and structured o…

Answer

TensorFlow is an open-source, end-to-end platform for Machine Learning, primarily developed by the Google Brain Team. It is used to build, train, and deploy machine learning models.

Answer

The name comes from its core mechanism: tensors, which are multi-dimensional arrays of data, flow through a graph of operations.

Answer

The five steps are: 1) Prepare Data, 2) Define Model, 3) Compile, 4) Fit or Train, and 5) Evaluate.

Answer

Data is loaded and reshaped or normalized so that the network can read it properly. For image data, pixel values are often scaled from 0 to 1 for better training stability.

Answer

The Sequential model is the easiest way to define a model architecture. It represents a linear stack of layers, where one layer follows another in order.

Answer

A Dense layer is a fully connected layer in which every neuron is connected to every neuron in the previous layer. It performs the heavy lifting of learning patterns.

Answer

A Flatten layer converts a 2D image into a 1D vector. For example, a 28x28 image becomes a line of 784 numbers, because neural networks generally need 1D vectors as input.