Lessons in Machine Learning Engineering – Part 4: Modeling | by David Martin | January, 2025

Practical ideas for a data-driven approach to model development

In this final part of my series, I'll share what I've learned in choosing an image classification model and how to tune that model. I will also show how you can use the model to speed up your labeling process, and finally how to justify your efforts by generating usage and performance statistics.

In Part 1, I discussed the process of labeling your image data that you use in your image classification project. Show me how to define “good” images and create sub-classes. In part 2, I went through the various datasets, beyond the typical test sets for train validation, and benchmark sets, and how to handle synthetic data and duplicate images. In Part 3, I explained how to apply different test criteria to a trained model versus a deployed model, and to use benchmarks to determine when the model should be deployed.

Model selection

So far I have focused most of my time on labeling and editing the image set, as well as testing the performance of the models, which is like putting the cart before the horse. I'm not trying to minimize what it takes to design a large neural network — this is the most important part of the program you're building. In my…