Class discussions of the Poisson distribution usually take second place to discussions of binomial probability distributions. Cognizant of this, the authors have conducted research on learning an unknown Poisson binomial distribution and present the results in these proceedings. Their purpose is to develop a highly efficient algorithm for dealing with this basic class of distributions independent of the population size.
After an introduction using the practical example of newspaper distribution effectiveness, the paper discusses several theorems and several applications of those theorems. These theorems and proofs are very sophisticated and not for beginners. There is a detailed appendix and reference section consisting of several pages.
In their conclusion, the authors state that they “have essentially settled the sample and time complexity of learning an unknown Poisson binomial distribution.”