From the course: Responsible AI and Application Development
Join today to access over 25,300 courses taught by industry experts.
Optimization: Responsible neural architecture design
From the course: Responsible AI and Application Development
Optimization: Responsible neural architecture design
- So far, we've been looking at data and its role in creating artificially intelligent applications. It's one of the most important parts of your system and one of the ways that bias and irresponsible practices can be introduced in a way that ultimately ruins your solution. But data is not the only potential source of problems. And a focus solely on data can be a naive one when you want to build responsibly. You should also focus on your neural network architecture. And while this is a very broad area, I would like to introduce you to some general concepts that you need to consider where biases can be introduced without you even being aware of them. The first one is optimization objectives. There's an old saying, "You are what you measure." And often when we look at neural networks and the training of neural networks, our goal is to be as accurate as possible. In particular with things like classification models, the common wisdom is that the more accurate the model, the better. A…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.