Built independently by an author, for readers. Read the story and support ChapterPal
ChapterPal Logo
Log inSign up

keyword

dataless classification

Dataless classification is a machine learning paradigm in which data instances, such as texts or documents, are categorized into predefined classes without relying on labeled training examples for the target task. Instead of learning statistical patterns from annotated datasets, this approach exploits the inherent semantic meaning of category names or descriptions by mapping both the input instances and the candidate labels into a shared semantic space. The model then assigns classes by measuring the semantic similarity or relevance between an input representation and the label representations. By substituting task-specific training data with semantic representations derived from general world knowledge, knowledge bases, or pre-trained language models, dataless classification allows systems to categorize inputs into new or evolving label sets on demand.

1 item

Few-Shot Learning with Siamese Networks and Label Tuning

Few-Shot Learning with Siamese Networks and Label Tuning

Thomas Müller, Guillermo Pérez-Torró, Marc Franco-Salvador

OrganizationsSymanto Research

Why you should read this

Introduces a scalable Siamese network framework and label tuning method for few-shot text classification that achieves constant-time inference and allows a single text encoder to be shared across multiple tasks by adapting only label embeddings.

We study the problem of building text classifiers with little or no training data, commonly known as zero and few-shot text classification. In recent years, an approach based on neural textual entailment models has been found to give strong results on a diverse range of tasks. In this work, we show that with proper pre-training, Siamese Networks that embed texts and labels offer a competitive alternative. These models allow for a large reduction in inference cost: constant in the number of labels rather than linear. Furthermore, we introduce label tuning, a simple and computationally efficient approach that allows to adapt the models in a few-shot setup by only changing the label embeddings. While giving lower performance than model fine-tuning, this approach has the architectural advantage that a single encoder can be shared by many different tasks.

Added

2026-10-03