Fine-tuned generative LLM oversampling can improve performance over traditional techniques on multiclass imbalanced text classification.
Nicolas Antonio CloutierNathalie JapkowiczPublished in: BigData (2023)
Keyphrases
- multi class
- text classification
- fine tuned
- feature selection
- binary classification problems
- single class
- support vector machine
- multiclass problems
- multiclass classification
- cost sensitive
- class imbalance
- binary classification
- pairwise
- multi label
- text categorization
- multi class classification
- fine tuning
- error correcting output codes
- multiple classes
- multi class problems
- multiclass support vector machines
- perceptron algorithm
- domain specific
- decision trees
- base classifiers
- naive bayes
- generative model
- general purpose
- support vector
- multiclass learning