How much training data do you really need?

After working with hundreds of companies deploying AI models in production, we've learned that ML teams often overestimate the amount of labelled data they think need to get a working model. 

On the other hand, teams also underestimate the amount of labeled data in order to get a first-class model. 

To share how teams get to good and then great, Labelbox lead machine learning engineer Randall Lin discusses how to think through the amount of training data needed for your projects, how to scale these efforts, and will take live questions from the audience.

In this webinar, you'll gain:

• A deeper understanding of the relationships between training set size, computational scale, and model accuracy improvements to advance state-of-the-art AI application development.

• An intuition for how much labeled data is necessary and at which points in your model development lifecycle this is necessary.