Organiza tus páginas con colecciones
Guarda y categoriza el contenido según tus preferencias.
Hasta ahora, este curso se enfocó en la creación de modelos de aprendizaje automático (AA).
Sin embargo, como se sugiere en la figura 1, los sistemas de AA de producción del mundo real son grandes
de ecosistemas nuevos y el modelo es solo una parte única y relativamente pequeña.
Figura 1: Un sistema de AA de producción real consta de muchos componentes.
El AA es el centro de un sistema de producción de aprendizaje automático del mundo real
pero a menudo representa solo el 5% o menos de la base de código total
el sistema. No es un error tipográfico; es significativamente menor de lo que
lo que espera. Ten en cuenta que un sistema de producción de AA dedica recursos considerables
a los datos de entrada: recopilarlos, verificarlos y extraer atributos.
[null,null,["Última actualización: 2025-07-27 (UTC)"],[[["\u003cp\u003eThis module explores the broader ecosystem of a production ML system, emphasizing that the model itself is only a small part of the overall system.\u003c/p\u003e\n"],["\u003cp\u003eYou will learn to choose the appropriate training and inference paradigms (static or dynamic) based on your specific needs.\u003c/p\u003e\n"],["\u003cp\u003eThe module covers key aspects of production ML systems, including testing, identifying potential flaws, and monitoring the system's components.\u003c/p\u003e\n"],["\u003cp\u003eAs a prerequisite, familiarity with foundational machine learning concepts, including linear regression, data types, and overfitting, is assumed.\u003c/p\u003e\n"],["\u003cp\u003eBuilding upon previous modules, this content shifts focus to the practical aspects of deploying and maintaining ML models in real-world scenarios.\u003c/p\u003e\n"]]],[],null,["# Production ML systems\n\n| **Estimated module length:** 70 minutes\n| **Learning objectives**\n|\n| - Appreciate the breadth of components in a production ML system.\n| - Pick the ideal training paradigm (static versus dynamic).\n| - Pick the ideal inference paradigm (static versus dynamic).\n| - Test your machine learning deployment.\n| - Ask the right questions about your production ML system.\n| - Determine flaws in real-world ML models.\n| - Monitor the components in a production ML system.\n| **Prerequisites:**\n|\n| This module assumes you are familiar with the concepts covered in the\n| following modules:\n|\n| - [Introduction to Machine Learning](/machine-learning/intro-to-ml)\n| - [Linear regression](/machine-learning/crash-course/linear-regression)\n| - [Working with numerical data](/machine-learning/crash-course/numerical-data)\n| - [Working with categorical data](/machine-learning/crash-course/categorical-data)\n| - [Datasets, generalization, and overfitting](/machine-learning/crash-course/overfitting)\n\nSo far, this course has focused on building machine learning (ML) models.\nHowever, as Figure 1 suggests, real-world production ML systems are large\necosystems and the model is just a single, relatively small part.\n**Figure 1.** A real-world production ML system comprises many components.\n\nAt the heart of a real-world machine learning production system is the ML\nmodel code, but it often represents only 5% or less of the total codebase in\nthe system. That's not a misprint; it's significantly less than you might\nexpect. Notice that an ML production system devotes considerable resources\nto the input data: collecting it, verifying it, and extracting features from it. \n[Help Center](https://support.google.com/machinelearningeducation)"]]
