Workshop F3: Smoothness and monotonicity constraints for neural networks using ICEnet
Deep neural networks have become an important tool for use in actuarial tasks. This is due to the significant gains in accuracy provided by these techniques compared to traditional methods and the close connection of these models to the generalised linear models (GLMs) used in industry.
In this work Old Mutual Insure:
- presents a novel method for enforcing constraints within deep neural network models
- shows how these models can be trained
- provides example applications using real-world datasets
Old Mutual Insure calls its proposed method ICEnet to emphasise the close link of its proposal to the individual conditional expectation (ICE) model interpretability technique.
Speaker: Ronald Richman, Old Mutual Insure