Asymptotic Risk of Overparameterized Likelihood Models: Double Descent Theory for Deep Neural Networks.
Ryumei NakadaMasaaki ImaizumiPublished in: CoRR (2021)
Keyphrases
- neural network
- neural network model
- probabilistic model
- experimental data
- computational models
- data sets
- parameter estimation
- complex systems
- pattern recognition
- neural nets
- mathematical framework
- theoretical framework
- statistical model
- prior knowledge
- decision making
- risk assessment
- learning rules
- high risk
- theoretical models
- expected values