Don’t Flood Your Algorithms With Easy Examples — It Costs Money | by Oscar Leo | Dec, 2024

Welcome back to a new ML Lesson for Managers and Engineers, where I share machine learning lessons drawn by mistakes and misconceptions I encounter across industries running my company, NextML! 🔥

Today, we’re looking at an error common among even the most experienced machine learning engineers and data scientists. I’ve seen it across industries, in large and small companies, and for a wide range of use cases.

The mistake is to flood algorithms with easy examples during training, leading to slower learning, worse generalization, and higher sensitivity to outliers.

Even more critical for most businesses, slow training of machine learning algorithms burns through your finances faster than necessary!

Note: In my experience, managers make bad decisions about their machine learning and AI strategy because they don’t understand the technology. I want to change that by providing lessons with a good balance between technical understanding and underlying reasoning.