Google has released a 60-minute training module designed to help machine-learning practitioners consider fairness as they develop machine-learning models. The module was created by Google’s engineering-education and machine-learning fairness teams and is part of the tech company’s Machine Learning Crash Course. The course teaches users on the different types of human biases that can appear in machine-learning models through data, the best ways to spot human biases in data before a model is trained, and methods to evaluate a model’s predictions for both over performance and bias.
Full Story