Mildly Overparametrized Neural Nets can Memorize Training Data Efficiently.
Rong GeRunzhe WangHaoyu ZhaoPublished in: CoRR (2019)
Keyphrases
- neural nets
- training data
- feed forward
- backpropagation neural networks
- multi layer
- back propagation
- artificial neural networks
- neural network
- counter propagation
- supervised learning
- decision trees
- data sets
- test set
- test data
- learning tasks
- single layer
- noisy data
- prior knowledge
- object recognition
- training examples
- learning algorithm
- genetic algorithm
- training process
- training samples
- real world
- knowledge discovery