IDoFew: Intermediate training using dual-clustering in language models for few labels text classification

Keyphrases

• Text Classification 100%
• Language Model 100%
• Few Labels 100%
• Dual Clustering 100%
• Clustering Algorithm 66%
• Pseudo-labeling 66%
• Copyright 33%
• Prediction Error 33%
• Error-prone 33%
• Text Mining 33%
• Natural Language Processing 33%
• Single-stage 33%
• Process Mining 33%
• Pre-trained Language Model 33%
• Natural Language Text 33%
• Bidirectional Encoder Representations from Transformers 33%
• Cold-start Problem 33%
• Training Step 33%
• Two-stage Clustering 33%
• Complementary Algorithm 33%
• Reliable Pseudo-labels 33%
• Limited Labels 33%

Computer Science

• Text Classification 100%
• Language Modeling 100%
• Clustering Algorithm 66%
• Text Mining 33%
• Natural Language Processing 33%
• Bidirectional Encoder Representations From Transformers 33%
• Pre-Trained Language Models 33%
• Cold Start Problem 33%
• Prediction Error 33%