What is Active Few-Shot Learning

Few-shot learning refers to the model’s ability to classify new data when only a limited number of training instances (e.g., 10 to 100) have been provided. As a result, after being exposed to a small amount of prior information, the model improves its performance.

Active learning is a type of machine learning where the model is trained on the data it considers most relevant, by prioritising the instances which are considered to be the most useful, and in that way reduces the data needs.

Few-shot learning and active learning combined together are known as active few-shot learning. Active few-shot learning aims at tackling the challenge of quickly developing and deploying new models for real-world problems with the least effort.

Open

Advantages

The Few-shot model has multiple advantages including:

No need for large annotated datasets. FASL models learn from few examples and thus require less data to be trained.

Reduced computational costs. As FASL models are trained with small data sets, this drastically reduces the costs related to data collection and the following manual annotation in-house or by third parties.

Increased model quality. Due to the combination of technology, in multiple use cases FASL surpasses the zero-shot model performance and allows you to reach a very high quality of your model.

Key elements

To create any few-shot model using Symanto Brain, there are two key elements required: 

• Classes or also called labels, between which to discriminate (classify) the text, e.g. NEGATIVE POSITIVE

• Patterns, also known as label descriptors, allow the semantic matching between the analysed text and the different labels, e.g. This text is {}

Examples

Please head over to Use Case Examples to view some common use cases.

Next: How to use Active Few-Shot Learning