Question

For a given hypothesis space (model space), adding more data can reduce the bias(approximation) error?

Fig: 1

Fig: 2