SPAKT: A Self-Supervised Pre-TrAining Method for Knowledge Tracing.
Yuling MaPeng HanHuiyan QiaoChaoran CuiYilong YinDehu YuPublished in: IEEE Access (2022)
Keyphrases
- prior knowledge
- significant improvement
- synthetic data
- clustering method
- training process
- preprocessing
- training samples
- high accuracy
- high precision
- cost function
- knowledge acquisition
- similarity measure
- neural network
- test data
- training phase
- classification method
- knowledge sharing
- error rate
- model selection
- medical images
- edge detection
- pairwise
- computational complexity
- feature extraction