Word embedding dimensionality reduction using dynamic variance thresholding (DyVaT).
Avraham TreistmanDror MughazAriel StulmanAmit DvirPublished in: Expert Syst. Appl. (2022)
Keyphrases
- dimensionality reduction
- structure preserving
- nonlinear dimensionality reduction
- high dimensional data
- graph embedding
- pattern recognition
- data representation
- high dimensional
- low dimensional
- locality preserving projections
- edge detection
- co occurrence
- feature extraction
- computer vision
- post processing
- embedding space
- low dimensional spaces
- word pairs
- neighborhood preserving
- correlation coefficient
- covariance matrix
- euclidean distance
- n gram
- data sets
- principal component analysis
- denoising
- multiscale
- image processing
- machine learning