Custom Classifier 2.0

How Our Custom Classifier 2.0 API Works?

Custom Classifier 2.0 works on the Zero-Shot Learning technique. This means, our base model is able to classify text it has not been trained on. It is achieved through transfer learning, by which knowledge learned from one dataset can be applied on a different one . People have proposed multiple zero shot learning approaches for vision tasks, but not text classification. This makes Custom Classifier 2.0 one of the earliest examples of applying zero-shot learning to text classification.

Customer Classifier 2.0 uses an algorithm that is trained to learn relationships between sentences and the labels they can be associated with. It is trained on a large corpus of data sets. Based on the paradigm. " Train Once, Test Anywhere", it has the ability to even generalize to new categories and on new datasets.

It also uses multiple neural network algorithms that use a training methodology to get near accurate results on a variety of datasets. One of the best methodologies is to use a Long Short-Term Memory (LSTM) model for the task of learning relationships. Through constant feedback connections, the LSTM algorithm used in Custom Classifier 2.0 helps in establishing the concept of "belongingness" between sentences and classes in the model. This knowledge is now useful for unseen classes or even unseen datasets.