On the class separability of contextual embeddings representations - or "The classifier does not matter when the (text) representation is so good!".
Cláudio Moisés Valiense de AndradeFabiano BelémWashington CunhaCelso FrançaFelipe ViegasLeonardo RochaMarcos André GonçalvesPublished in: Inf. Process. Manag. (2023)
Keyphrases
- class separability
- text representation
- feature selection
- feature space
- dimensionality reduction
- intra class
- text classification
- text categorization
- information filtering
- linear discriminant analysis
- discriminant analysis
- concept learning
- low dimensional
- document representation
- text documents
- bag of words
- contextual information
- training data
- index terms
- support vector machine
- keywords
- vector space
- feature set
- training set
- vector space model
- high dimensional data
- feature extraction
- document clustering
- image processing
- decision trees
- information extraction
- image representation
- high dimensional
- knn
- machine learning
- information retrieval
- support vector
- text retrieval
- visual words
- distance measure
- input data
- neural network
- semi supervised learning