Researchers have developed ICARE, a new framework designed to improve the performance of natural language inference (NLI) systems, particularly in cross-domain text classification. The framework addresses the issue of catastrophic forgetting, where NLI systems lose previously learned knowledge when trained on new tasks. ICARE utilizes an instance selection module based on representation learning to choose optimal training examples and a knowledge distillation method that transfers knowledge at both feature and logit levels. Evaluations on benchmark datasets show that ICARE significantly enhances target accuracy and reduces forgetting, making NLI systems more adaptable and effective across diverse tasks. AI
IMPACT Enhances adaptability and performance of NLI systems in cross-domain text classification tasks.
RANK_REASON The cluster contains a research paper detailing a new framework for NLI systems. [lever_c_demoted from research: ic=1 ai=1.0]
- arXiv
- Coreference Resolution
- Hugging Face
- ICARE
- knowledge distillation
- News Classification from Social Media Using Twitter-based Doc2Vec Model and Automatic Query Expansion
- sentiment analysis
- text classification
- natural language inference
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →