Governing Deep Learning Models

Deploying a model is just the beginning of its journey. Once it’s out in the real world, it’s like a living thing – it requires efficient use to make the most of it, upgrades to stay sharp, care to perform consistently well, and, eventually, a graceful exit. Imagine a car on the road: you start driving, but you also need to use the car effectively, fuel it, maintain it, and eventually replace it or its components. The same goes for deep learning models in action.

Model governance acts as the guiding force that oversees the use of a model and maintains constant vigilance over its performance and context to ensure the continuous, consistent, and dependable delivery of value through the model. In the realm of deep learning, model governance is crucial for ensuring that these complex models adhere to the highest standards of quality, reliability, and fairness.

This chapter delves into the three fundamental pillars of model governance...