You have an Azure Machine Learning workspace. You plan to use the workspace to set up automated machine learning training for an image classification model. You need to choose the primary metric to…

Image classification is a multiclass classification task where the goal is to correctly assign each image to a category, and accuracy - the ratio of correctly predicted labels to total samples - is the standard metric for this objective. Azure AutoML for image tasks natively supports accuracy as a primary metric and maximizes it during model selection and hyperparameter optimization.

Mean absolute error measures the average magnitude of prediction error for continuous numerical outputs and is a regression metric not applicable to classification problems.

Spearman correlation measures the monotonic rank relationship between predicted and actual continuous values and is used in regression tasks, not image classification.

Root mean squared log error penalizes underestimates of continuous numeric predictions and is a regression metric with no meaningful application to image classification tasks.