Backend

As explained in Deep Learning with Python, Third Edition

The underlying execution framework that a machine-learning library can target to run models.

A backend is the low-level execution framework beneath Keras. It supplies tensor operations, gradients, and hardware access while Keras provides the higher-level model and training APIs. section 3.6.1 section 3.2

Selecting a Keras backend

Choose the backend before importing Keras. For example, this selects JAX for the current Python process: import os os.environ["KERAS_BACKEND"] = "jax" import keras You can also edit ~/.keras/keras.json. Use "torch" for the PyTorch backend, not "pytorch". A fresh Keras environment uses TensorFlow by default. section 3.6.1

Use these practical rules when choosing or changing a backend:

• Switch among JAX, TensorFlow, and PyTorch without locking Keras code to one ecosystem. section 3.6.1
• Match the backend to the task: debugging, TPU training, or inference tooling can favor different ecosystems. section 3.6.1
• Treat NumPy differently: it supports Keras predictions but cannot support training because it provides no gradient API. section 3.2

Figure 3.4 Keras and its backends. A backend is a low-level tensor-computing platform; Keras is a high-level deep learning API.

Use the diagram to distinguish Keras’s high-level API from the low-level tensor-computing platform supplied by its backend. section 3.6.1

How Deep Learning with Python, Third Edition covers it

How deep it goes

Explains backend selection, portability, tensor operations, and gradients, while stopping short of framework-specific implementation details.

Examples

Uses KERAS_BACKEND, ~/.keras/keras.json, JAX selection, and NumPy’s prediction-only limitation as concrete examples.

What it assumes

Assumes familiarity with Keras at a basic level while distinguishing its high-level APIs from backend-provided tensors and gradients.

What you can do after

Operate/configure Keras by selecting a backend before import and matching it to debugging, training, or inference needs.

How it connects

Connects Keras with TensorFlow, PyTorch, JAX, NumPy, tensors, gradients, hardware, and training loops.

liveBookmanning.com

Read Deep Learning with Python, Third Edition on liveBook Get the book at manning.com

Also explained in

• 

  liveBookmanning.com
  Rust Servers, Services, and AppsA backend service is a server-side application component that implements application functionality and exposes it to clients or other services.
• 

  liveBookmanning.com
  Building User-Friendly DSLsThe server-side part of a software system that provides services, storage, and processing for its frontend.
• 

  liveBookmanning.com
  Deep Learning with R, Third EditionA backend engine is a software system that executes the numerical and machine-learning operations required by a higher-level framework.

Compare how 4 books explain backend →

Related concepts

• web applicationpart of · A web application is software that provides functionality through web technologies and is accessed by users through a browser or other web client.
• databaseuses · A database is an organized collection of persistent data managed by software for storage, retrieval, querying, and modification.
• natural language interfaceused by · A natural language interface enables interaction with a system through human-language text or speech.
• frontend developmentcontrasts with · Frontend development is the practice of building the user-facing, client-side interfaces and behavior through which people interact with software.
• JAXincludes · JAX is a Python library for high-performance array-based numerical computing with composable transformations including automatic differentiation, compilation, vectorization, and parallelization.
• contractuses · An API contract is a specification of the operations, data, and behavioral expectations that govern interactions between software components.

This explainer was generated by Manning's knowledge graph from the text of Deep Learning with Python, Third Edition. The section links point to the parts of the book it draws on.