- TensorFlow is an open-source machine learning framework for performing machine and deep learning tasks using data flow graphs and numerical computation. It can run on CPUs, GPUs, and TPUs.
- Key concepts include tensors (multi-dimensional arrays), flow graphs representing operations as nodes and tensors as edges, and sessions for executing graphs. TensorFlow 2.0 fully integrates the Keras API.
- TensorFlow is used for tasks like computer vision, natural language processing, speech recognition through neural networks and deep learning models. Popular datasets include MNIST for images and IMDB for text.