On the class separability of contextual embeddings representations – or “The classifier does not matter when the (text) representation is so good!”
Authors:
- Claudio M.V. de AndradeCAClaudio M.V. de Andrade
- Fabiano M. BelémFBFabiano M. Belém
- Washington CunhaWCWashington Cunha
- Celso FrançaCFCelso França
- Felipe ViegasFVFelipe Viegas
- Leonardo RochaLRLeonardo Rocha
- Marcos André GonçalvesMGMarcos André Gonçalves
- 13 April 2023
- journal article
- research article
- Published by Elsevier in Information Processing & Management
- Vol. 60 (4)
Abstract
No abstract availableKeywords
This publication has 14 references indexed in Scilit:
- On the cost-effectiveness of neural and non-neural approaches and representations for text classification: A comprehensive comparative studyInformation Processing & Management, 2021
- Deep Learning--based Text ClassificationACM Computing Surveys, 2021
- Extended pre-processing pipeline for text classification: On the role of meta-feature representations, sparsification and selective samplingInformation Processing & Management, 2020
- TANDA: Transfer and Adapt Pre-Trained Transformer Models for Answer Sentence SelectionProceedings of the AAAI Conference on Artificial Intelligence, 2020
- Enriching Word Vectors with Subword InformationTransactions of the Association for Computational Linguistics, 2017
- Turning from TF-IDF to TF-IGM for term weighting in text classificationExpert Systems with Applications, 2016
- Word co-occurrence features for text classificationInformation Systems, 2011
- Margin based feature selection - theory and algorithms2004
- Indexing by latent semantic analysisJournal of the American Society for Information Science, 1990
- Silhouettes: A graphical aid to the interpretation and validation of cluster analysisJournal of Computational and Applied Mathematics, 1987