You have deployed multiple versions of an image classification model on AI Platform.
You want to monitor the performance of the model versions over time.
How should you perform this comparison?
Click on the arrows to vote for the correct answer
A. B. C. D.B.
To monitor the performance of multiple versions of an image classification model over time, the recommended approach is to use Continuous Evaluation, which is a feature of AI Platform.
Continuous Evaluation enables the comparison of model performance across different versions, by automatically evaluating the model on a scheduled basis and calculating evaluation metrics, such as mean average precision (MAP).
Therefore, the correct answer to this question is option D: Compare the mean average precision across the models using the Continuous Evaluation feature.
Option A and B are incorrect because comparing the loss performance on a held-out dataset or validation data does not provide a comprehensive view of the model's performance. These metrics only measure the model's ability to minimize the loss on the data it has seen before and may not reflect its performance on new data.
Option C is also incorrect because ROC curves are used to evaluate binary classification models, and it is not clear whether the image classification model is binary or multi-class.
In conclusion, the recommended approach to compare the performance of multiple versions of an image classification model is to use Continuous Evaluation and measure mean average precision.