Video by freeCodeCamp.org · taught by Tim Ruscica · streamed from YouTube. All rights remain with the creator.
Lesson 2: TensorFlow 2 Crash Course
About this lesson
This crash course by Tim Ruscica teaches you to build neural networks with Python and TensorFlow 2 through short, practical projects, and is a good starting point before the longer complete course. It opens by explaining what a neural network is, how layers of connected neurons turn inputs into outputs and how training adjusts their weights and biases so that the predictions improve. The course then shows how to load a dataset and look at it before doing anything else, checking the shapes of the arrays and displaying sample images with their labels, and how to prepare the values for training. Next it creates a model with Keras, choosing the layers, activation functions, loss function and optimiser, then trains it and measures its accuracy on test data that it has not seen. With a trained model in hand, the course demonstrates how to use it to make predictions on new examples and how to interpret its outputs. The second project moves from images to text, classifying film reviews as positive or negative. It covers how the reviews are encoded as integers, how they are padded to a common length and how a network learns word embeddings, as well as how to save a trained model and load it again later rather than retraining it. The course ends by explaining how to install the graphics processing unit version of TensorFlow on Linux for faster training. Afterwards, read Chapter 2 of the course book, which counts the parameters of a text model by hand.