TensorFlow Basics

Study Notes

An open-source software library for numerical computation, particularly well-suited for large-scale machine learning and deep learning tasks.
Developed by the Google Brain team, initially released in 2015.

Automatic Differentiation: computes gradients of the loss function with respect to the model's parameters.
Distributed Training: allows training of models on multiple machines, making it scalable.
AutoGraph: converts Python code into TensorFlow graphs, enabling more efficient execution.
Eager Execution: enables immediate execution of TensorFlow operations, making it easier to debug and test.

TensorFlow Graph: a dataflow graph that represents the computation, consisting of nodes (operations) and edges (data dependencies).
TensorFlow Session: a runtime environment that executes the graph, providing a way to interact with the graph.
TensorFlow Ops: the basic operations that can be performed on tensors, such as addition, multiplication, and matrix multiplication.

Tensors: multi-dimensional arrays of numerical values, used to represent data and model parameters.
Tensor Ranks: the number of dimensions in a tensor, with scalar (0), vector (1), matrix (2), and tensor (3) being common examples.
Tensor Shapes: the size of each dimension in a tensor, e.g., a 3x4 matrix has a shape of (3, 4).

TensorFlow Core: the low-level API, providing direct access to TensorFlow's functionality.
Keras API: a high-level API, built on top of TensorFlow, providing an easier-to-use interface for building and training machine learning models.
Estimators API: a high-level API, providing a simpler way to train and evaluate machine learning models.

Machine Learning: TensorFlow is widely used for building and training machine learning models, including neural networks, decision trees, and support vector machines.
Deep Learning: TensorFlow is particularly well-suited for deep learning tasks, such as image and speech recognition, natural language processing, and generative models.
Computer Vision: TensorFlow is used in computer vision applications, such as object detection, image segmentation, and image generation.