Summary: Many problems in combinatorics, number theory, probability theory, reliability theory and statistics can be solved by applying a unifying method, which is known as the principle of inclusion-exclusion. The principle of inclusion-exclusion expresses the indicator function of a union of finitely many events as an alternating sum of indicator functions of their intersections.
This thesis deals with improved inclusion-exclusion identities and improved Bonferroni inequalities that require the family of events to satisfy some structural restrictions. Examples of such well-structured families arise in problems of statistical inference, combinatorial reliability theory and chromatic graph theory.