About Machine Learning Studio

Machine Learning Studio uses deep learning models to enable document understanding. With ML Studio, you annotate sample documents and evaluate and train deep learning models.

The models you develop in ML Studio form a key piece of most Instabase solutions. Models are responsible for classifying documents and extracting data.

Model development begins with annotation, which demonstrates how you want documents to be classified and what data you want extracted. You then use the annotation set to evaluate and train models. If your use case involves common document types, you might be able to leverage Marketplace models to skip or fast-track model training.

Models address a specific use case, such as classifying various types of IDs, but you can combine multiple models in a single flow or solution. Similarly, you can use the same annotation set to evaluate and train multiple models.

You can approach annotation and model training sequentially or iteratively. In an iterative approach, you annotate at least the minimum number of documents, train a model, and then use your model training results to refine or expand your annotation set.

Both annotation and model training might be handled by a solution developer or, for large or complex annotation sets, annotation might be handed off to dedicated annotators.

Much of the functionality in ML Studio is provided through training scripts, in the form of an ibformers package, and base models. Both of these dependencies are provided through the Marketplace. Before you begin working on a new ML Studio project, make sure you have the latest dependencies by updating Marketplace.