All Products
Search
Document Center

Platform For AI:AutoML

Last Updated:Mar 18, 2024

Platform for AI (PAI) provides AutoML to help you search for optimal hyperparameter combinations based on specific policies. You can use AutoML to improve the efficiency of model tuning. This topic provides an overview of AutoML and links to related topics.

What is AutoML?

AutoML is an enhanced machine learning service provided by PAI to help you find optimal hyperparameter combinations. In model training scenarios in which the hyperparameter combination is complex and requires a large amount of training resources and fine-tuning workload, you can use AutoML to fine-tune the hyperparameters in an efficient manner and improve model quality.

How AutoML works

AutoML automates hyperparameter fine-tuning by iterating experiments, trials, and training tasks to find the optimal hyperparameter combination.

Limits and usage notes of AutoML

This topic describes the limits and usage notes of AutoML, including the supported regions, search algorithms, and suitable scenarios. The supported search algorithms include Tree-structured Parzen Estimator (TPE), grid search, random search, evolutionary algorithm, Gaussian Processes (GP), and population based training (PBT).

Grant the permissions that are required to use AutoML

Before you use AutoML, you need to activate the Alibaba Cloud services on which AutoML depends, such as Deep Learning Containers (DLC) of PAI, MaxCompute, and Object Storage Service (OSS).

Console operations

Create an experiment

AutoML is an enhanced machine learning service provided by PAI that integrates multiple algorithms and distributed computing resources. AutoML allows you to fine-tune model hyperparameters in a more efficient manner by creating experiments without coding. You can also use AutoML to improve model performance. This topic describes how to create an experiment.

View experiment details

After you create an experiment, you can view the experiment details in real time, such as the basic information about experiments and trials and the execution and log details of each trial. This topic describes how to view the details of an experiment.

Manage experiments

After you create an experiment, you can manage the experiment and trials of the experiment, such as cloning, modifying, stopping, and deleting an experiment or a trial. This topic describes how to manage experiments.

AutoML use cases

This topic provides links to AutoML use cases.