Mitigating Dataset Artifacts in Natural Language Inference Through Automatic Contextual Data Augmentation and Learning Optimization.
Michail MersiniasPanagiotis ValvisPublished in: LREC (2022)
Keyphrases
- data sets
- natural language
- data processing
- prior knowledge
- data analysis
- learning models
- learning process
- data sources
- context sensitive
- high quality
- database
- data points
- learning tasks
- learning systems
- high dimensional data
- data collection
- semantic analysis
- data quality
- original data
- structured prediction
- input data
- machine learning
- training data
- knowledge discovery
- supervised learning
- optimization problems
- online learning
- question answering
- missing data
- background knowledge
- experimental data
- raw data
- knowledge representation
- evolutionary algorithm
- optimization algorithm
- e learning
- bayesian methods
- neural network