This topic provides an example of using AutoLearning to train models. In this example, an image classification model is trained to classify images of mules, horses, and alpacas.

Overview

The system provides 33 images of mules, horses, and alpacas. You can use these images to train a model with an accuracy of higher than 80%, then use the model to classify mules, horses, and alpacas. After you upload an image into the model, the category of the image is returned.

Procedure

  1. Perform the following steps to navigate to the AutoLearning page.
    1. Log on to the Machine Learning Platform for AI console.
    2. In the left-side navigation pane, choose AutoLearning > General Purpose Model Training.
  2. On the AutoLearning page, navigate to the Image Classification_Animal section, and click Create.
  3. Label images.
    1. In the Data Annotation wizard, navigate to the List of labels section, and click Add Label.
    2. In the text box that appears, enter mule. Then, press the Enter key.
    3. Repeat the preceding steps to add labels of horse and alpaca.
    4. Select all images of alpacas. Then, click Select picture label, and select alpaca.Image labeling
    5. Repeat the preceding steps to label the images of horses and mules.
    6. Click Complete labeling.
  4. Train and evaluate the model.
    1. In the Start training dialog box, set the Maximum training duration to 10.
      AutoLearning supports Early Stopping, which can prevent overfitting.
      Note The Maximum training duration is measured in minutes. Enter a value from 10 to 60.
    2. Click Start training.
    3. In the Model Training and Evaluation wizard, view the Training progress and Latest Assessment Results.
      To increase the accuracy of the model, you can increase the number of images or training duration.
  5. Test the model.
    1. In the Model Training and Evaluation wizard, navigate to the Operation section, and click Trial.
    2. In the Model Trial wizard, drag a local image to the Local files section.
    3. Click Forecast.
    4. On the List Display tab, view the Confidence level of the prediction result.