On the Sensitivity of LSTMs to Hyperparameters and Word Embeddings in the Context of Sentiment Analysis.
Bousselham El HaddaouiRaddouane ChihebRdouan FaiziAbdellatif El AfiaPublished in: BDIoT (2021)
Keyphrases
- sentiment analysis
- sentence level
- hyperparameters
- opinion mining
- sentiment classification
- text classification
- model selection
- cross validation
- sentiment words
- opinion words
- gaussian process
- em algorithm
- maximum likelihood
- sample size
- positive or negative
- bayesian framework
- text mining
- natural language processing
- bayesian inference
- gaussian processes
- product features
- random sampling
- support vector
- learning algorithm
- artificial intelligence
- high dimensional
- support vector machine
- incomplete data
- prior information
- maximum a posteriori
- closed form
- dimensionality reduction