Mike Bishop comments on Mathematical simplicity bias and exponential functions

[Mike Bishop]

Yes, this is exactly the point I was getting at when I wrote: “Of course it is best to test the model on new data.”