Berry-esseen type theorems for sequential density estimation

Given a sequence of i.i.d. random variables the convergence of kernel-type sequential estimators of their common density function is studied. This throws light on the usefulness of such estimators. The results are also extended to the case when the sample size is random.