Learn the key steps for comparing multiple AI models trained on the same data to identify construction defects. Discover how to generate a confusion matrix to select the best-performing model.
Table of Contents
Question
Your AI team has developed several models to visually spot defects in building construction and trained them all on the same data. Now they’re about to generate a confusion matrix to select the best one.
What steps will this involve?
A. Run them all on different data sets of data, average their outcomes, then compare the results.
B. Run them all on the identical batches of new data, then compare their performance.
C. Run three out of four model on new data, then compare them with the fourth as a control group.
D. Run them all and time how long it takes each one to create an accurate prediction.
Answer
B. Run them all on the identical batches of new data, then compare their performance.
Explanation
You’ll run your models competitively, testing then all on the same new data set, then compare their results using a confusion matrix.
To select the best model out of several that were all trained on the same data:
- Prepare a new, previously unseen dataset that the models have not been trained on. This allows you to evaluate their real-world performance and generalization ability.
- Run each of the trained models on the exact same new dataset, generating predictions from every model on the identical data points. This ensures a fair, apples-to-apples comparison.
- For each model, generate a confusion matrix summarizing its performance on the new data. The confusion matrix shows the models’ true positives, true negatives, false positives, and false negatives.
- Compare the confusion matrix metrics across the models, looking at key indicators like accuracy, precision, recall, and F1 score. The model with the best overall performance metrics on the new data is the top candidate.
The other answer choices are incorrect:
A) Running the models on different datasets would not allow direct comparison of their performance.
C) Using one model as a control group is unnecessary; they should all be evaluated equally.
D) Prediction speed is a secondary concern compared to a model’s accuracy and performance.
By running all models on the same new data and comparing their confusion matrices, the AI team can confidently select the model that generalizes best to real-world construction defect identification.
IBM Artificial Intelligence Fundamentals certification exam practice question and answer (Q&A) dump with detail explanation and reference available free, helpful to pass the Artificial Intelligence Fundamentals graded quizzes and final assessments, earn IBM Artificial Intelligence Fundamentals digital credential and badge.