Multi-teacher knowledge distillation as an effective method for compressing ensembles of neural networks.
Konrad ZuchniakPublished in: CoRR (2023)
Keyphrases
- neural network
- preprocessing
- significant improvement
- high precision
- cost function
- genetic algorithm
- neural nets
- clustering method
- high accuracy
- fuzzy logic
- prior knowledge
- pairwise
- experimental evaluation
- computational cost
- probabilistic model
- input data
- computationally efficient
- training samples
- computational complexity
- similarity measure
- machine learning methods
- feature extraction